1 Orbital Evidence for Clay and Acidic Sulfate Assemblages on Mars Based on 2 Mineralogical Analogs from Rio Tinto, Spain 3 4 Hannah H. Kaplan1*, Ralph E. Milliken1, David Fernández-Remolar2, Ricardo Amils3, 5 Kevin Robertson1, and Andrew H. Knoll4 6 7 Authors: 8 9 1 Department of Earth, Environmental and Planetary Sciences, Brown University, 10 Providence, RI, 02912 USA. 11 12 2British Geological Survey, Nicker Hill, Keyworth, NG12 5GG, UK. 13 14 3 Centro de Astrobiologia (INTA-CSIC), Ctra Ajalvir km 4, Torrejon de Ardoz, 28850, 15 Spain. 16 17 4 Department of Organismic and Evolutionary Biology, Harvard University, 18 Çambridge, MA, USA. 19 20 *Corresponding author: [email protected]
1 21 Abstract 22 23 Outcrops of hydrated minerals are widespread across the surface of Mars,
24 with clay minerals and sulfates being commonly identified phases. Orbitally-based
25 reflectance spectra are often used to classify these hydrated components in terms of
26 a single mineralogy, although most surfaces likely contain multiple minerals that
27 have the potential to record local geochemical conditions and processes.
28 Reflectance spectra for previously identified deposits in Ius and Melas Chasma
29 within the Valles Marineris, Mars, exhibit an enigmatic feature with two distinct
30 absorptions between 2.2 – 2.3 µm. This spectral ‘doublet’ feature is proposed to
31 result from a mixture of hydrated minerals, although the identity of the minerals has
32 remained ambiguous. Here we demonstrate that similar spectral doublet features
33 are observed in airborne, field, and laboratory reflectance spectra of rock and
34 sediment samples from Rio Tinto, Spain. Combined visible-near infrared reflectance
35 spectra and X-ray diffraction measurements of these samples reveals that the
36 doublet feature arises from a mixture of Al-phyllosilicate (illite or muscovite) and
37 jarosite. Analyses of orbital data from the Compact Reconnaissance Imaging
38 Spectrometer for Mars (CRISM) shows that the martian spectral equivalents are also
39 consistent with mixtures of Al-phyllosilicates and jarosite, where the Al-
40 phyllosilicate may also include kaolinite and/or halloysite. A case study for a region
41 within Ius Chasma demonstrates that the relative proportions of the Al-
42 phyllosilicate(s) and jarosite vary within one stratigraphic unit as well as between
43 stratigraphic units. The former observation suggests that the jarosite may be a
44 diagenetic (authigenic) product and thus indicative of local pH and redox conditions,
2 45 whereas the latter observation may be consistent with variations in sediment flux
46 and/or fluid chemistry during sediment deposition.
47
48 Key points: 49 • A doublet absorption is observed in reflectance spectra of Mars between 2.2 - 50 2.3 µm. 51 • Reflectance spectra at Rio Tinto, Spain also show a doublet due to Al- 52 phyllosilicate and jarosite. 53 • Spectral variation in martian deposits suggests local variation in pH, redox, 54 and/or sediment flux. 55 56 57 Keywords: Mars, surface; Spectroscopy; Infrared observations 58
3 59 1. Introduction
60 Outcrops of sulfate and clay mineral-bearing rocks are widespread across
61 the ancient surface of Mars [Bibring et al., 2006; Murchie et al., 2009; Ehlmann et al.,
62 2011; Carter et al., 2013], indicating that a complex history of water-rock interaction
63 is preserved in the martian rock record. Visible-near infrared (VIS-NIR) reflectance
64 spectra acquired by the CRISM (Compact Reconnaissance Imaging Spectrometer for
65 Mars) and OMEGA (Observatoire pour la Minéralogie, l’Eau, les Glaces et l’Activité)
66 imaging spectrometers have been used to identify clay minerals in ancient and
67 heavily cratered Noachian terrains [Poulet et al., 2005; Ehlmann et al., 2011; Carter
68 et al., 2013] and thick deposits of sulfates in younger Hesperian terrains and basins
69 [Gendrin et al., 2005; Bibring et al., 2006]. In the cases of Terra Meridiani, Gusev
70 Crater, and Gale Crater, these minerals have been examined in situ by rover-based
71 payloads, providing detailed local geologic context for some of these orbital
72 detections [e.g. Klingelhofer et al., 2004; Squyres et al., 2004; Arvidson et al., 2008;
73 Wang et al., 2006; Vaniman et al., 2014].
74
75 Many of these orbiter-based detections of hydrated minerals on Mars are
76 discussed in terms of single minerals. That is, the reflectance spectra of these
77 deposits are commonly discussed or interpreted as specific clay, sulfate, or other
78 hydrated minerals (see Ehlmann et al., 2009 for an exception with regards to
79 mineral assemblages at Nili Fossae). Though these deposits are undoubtedly
80 polymineralic, it appears that in many cases a single mineral can be spectrally
81 dominant even at the spatial resolution of instruments such as CRISM (e.g., ~18
4 82 m/pixel). Therefore, those regions on Mars whose reflectance spectra cannot clearly
83 be attributed to a single dominant hydrated mineral are of particular interest as
84 they may record different and more diverse geological processes than are typically
85 recognized from CRISM data.
86
87 Reflectance spectra of deposits and strata within the Valles Marineris reveal
88 a number of locations that fall within this category, and these are the focus of this
89 study. Specifically, CRISM spectra of various deposits within Ius, Melas, and other
90 chasmata exhibit an absorption feature with local minima centered at ~2.21 and
91 ~2.27 μm [Metz et al., 2009; Roach et al., 2010; Weitz et al., 2011] (Figure 1). The
92 spectral shape and properties of this absorption feature, which we will refer to as a
93 spectral ‘doublet’, does not match the spectrum of any single common mineral in
94 existing spectral libraries, suggesting it likely cannot be attributed to a single type of
95 clay, sulfate, or other common hydrated mineral.
96
97 Previous studies have hypothesized that this material represents a mixture
98 of clays, sulfates (possibly including jarosite), and/or poorly crystalline phases
99 [Roach et al., 2010; Tosca et al., 2008b; Noe Dobrea et al., 2011; Weitz et al., 2014],
100 but the exact assemblage remains unknown. We have observed that similar spectral
101 features occur in airborne, field, and lab spectra of rocks and sediments at Rio Tinto,
102 Spain, which has been previously studied as a Mars analog due to the river’s acidity,
103 iron hydrochemistry, and extensive iron oxide deposits [Fernandez-Remolar et al.
104 2003]. Indeed, it has been suggested that the ferric iron and sulfate mineralogy
5 105 produced by the acidic waters may be analogous to conditions experienced in Terra
106 Meridiani during the Hesperian [Fernandez-Remolar et al., 2004], and similar
107 conditions may also have occurred in portions of the Valles Marineris [e.g., Bibring
108 et al., 2007].
109
110 In this study we integrate airborne, in situ, and lab measurements to
111 characterize a suite of samples from different sites at Rio Tinto to understand the
112 mineralogical origin of the ‘doublet’ spectral signature, and we then compare our
113 findings to reflectance spectra acquired by the CRISM instrument for deposits in
114 Valles Marineris, Mars. We demonstrate that the doublet absorptions observed in
115 samples from Rio Tinto are comparable to the spectral signatures observed in Valles
116 Marineris, that the former are the result of Al-bearing phyllosilicates mixed with
117 jarosite, and that similar mineral assemblages may be present on Mars. This
118 indicates Rio Tinto contains materials that can serve as mineralogical and spectral
119 analogs for the martian deposits, though the geological processes that lead to the
120 formation of these assemblages on Earth and Mars need not be equivalent.
121 Specifically, although jarosite and Al-phyllosilicates may co-occur in geologic units
122 today on Mars, this association does not require their formation environments to
123 have been the same in space and/or time. Both mineral types may be authigenic or
124 detrital, and detailed geologic context is necessary to distinguish between these
125 options and the associated implications for local paleoenvironmental conditions.
126
127 2. Background
6 128 2.1 Overview of Clay and Sulfate Detections on Mars
129 The most commonly observed spectral signatures on Mars that have been
130 interpreted as clay minerals are identified as Fe/Mg smectite, but mixed-layer clays,
131 chlorite, Al-smectite, mica, kaolinite, and serpentine have also been observed
132 [Mustard et al., 2008; Murchie et al., 2009; Ehlmann et al., 2011; Carter et al., 2013].
133 Aqueous alteration of basaltic protoliths both at the surface and in the subsurface
134 have been invoked as the major source of these clay minerals, although subsequent
135 erosion and eolian/fluvial transport has led to their occurrence in a number of
136 different depositional environments. Intriguingly, these clay minerals, hereafter
137 referred to simply as ‘clays’, occur predominantly in terrains that are Noachian
138 (>3.6 Ga) in age [Bibring et al., 2006]. The presence and apparent spatial and
139 temporal distribution of these clays has been used to infer that early Mars was
140 characterized by circum-neutral pH, low salinity, and possibly habitable conditions
141 [Ehlmann et al., 2011]. More recently, in situ observations of clay-bearing
142 mudstones in Gale Crater by the Curiosity rover are also supportive of a habitable
143 environment, possibly during the Hesperian [Grotzinger et al., 2014; 2015; Vaniman
144 et al., 2014; Bristow et al., 2015].
145
146 In contrast, sulfate detections on Mars typically occur in younger Hesperian
147 terrains and may be indicative of a decreased hydrological cycle [Bibring et al.,
148 2006]. Decreased water availability, coupled with increasing salinity, may indicate
149 near-surface conditions inhospitable to most life [e.g., Tosca et al., 2008a]. Though
150 many sulfate salts (e.g., Ca, Na, and Mg-sulfates) can form under a wide range of pH
7 151 conditions, it is worth noting that Al3+ and Fe3+-sulfates are indicative of low-pH
152 fluids. As such, understanding the distribution and geologic setting of these minerals
153 on Mars has important implications for deciphering ancient aqueous chemistry,
154 whether it be indicative of surface waters, groundwater, or diagenetic fluids.
155
156 Jarosite, a ferric sulfate that forms under acidic conditions and that was first
157 detected on Mars by the Opportunity Rover in Meridiani Planum [Squyres et al.,
158 2004], has received considerable attention in this context. Integrated payload
159 observations by the rover reveal that jarosite is likely associated with Mg and Ca-
160 sulfates at this location, all of which occur in eolian sandstones that have been
161 interpreted as reworked dune and playa sediments [Grotzinger et al., 2005]. The
162 Curiosity rover has also detected small amounts of jarosite in clay-bearing
163 laminated mudstones in Gale Crater [Cavanagh et al., 2015], though the origin of this
164 phase (e.g., detrital or authigenic) in these deposits is not yet well constrained.
165
166 Orbital observations of jarosite and other possible ferric sulfates have also
167 been reported in the Mawrth Vallis, Aram Chaos, Noctis Labyrinthus, and Valles
168 Marineris regions of Mars [Bibring et al., 2007; Milliken et al., 2008; Farrand et al.,
169 2009; Metz et al., 2009; Lichtenberg et al., 2010; Weitz et al., 2010; Thollot et al.,
170 2011; Noe Dobrea et al., 2011]. Of the Valles Marineris detections, possible
171 submarine fan deposits in Melas Chasma and strata exposed on the surrounding
172 plains are notable in that ferric sulfates are found in proximity to other hydrated
173 minerals such as opaline silica, providing additional constraints on local aqueous
8 174 geochemistry [Milliken et al., 2008; Metz et al., 2009]. These rover- and orbiter-
175 based observations indicate ferric sulfates may be common alteration or diagenetic
176 phases on Mars, as discussed by Burns [1988; 1993], and understanding their
177 occurrence can provide insight into local, regional, or global redox conditions.
178
179 2.2 CRISM Observations of the Spectral ‘Doublet’
180
181 CRISM spectra of martian deposits that exhibit the ‘doublet’ feature
182 mentioned above are characterized by absorption maxima (reflectance minima)
183 occurring at wavelengths of 2.205-2.218 and 2.265-2.278 μm. These spectral
184 features have been noted previously in a number of CRISM-based studies,
185 particularly for deposits within Valles Marineris, and in some of these studies the
186 material has been referred to as a hydrated silicate [e.g., Roach et al., 2010].
187 However, many minerals, including gypsum, jarosite, nontronite, and silica, exhibit
188 absorptions in this wavelength range, and it is possible that the spectral doublet
189 represents a mixture of two or more hydrated minerals (Figure 1). The main
190 doublet feature is consistent with vibrational absorptions due to the presence of
191 cation-OH bonds, although the position and shape of absorptions within the feature
192 lead to ambiguity in determining the specific cation(s). The shorter wavelength
193 absorption (~2.21 µm) may be consistent with Si- or Al-OH bonds, whereas the
194 longer wavelength feature (~2.27 µm) may be indicative of Fe- or Mg-OH bonds.
195 However, spectra of H2O-bearing Ca-sulfates such as gypsum also exhibit
9 196 absorptions in this wavelength region, despite the lack of structural Si, Al, Fe, Mg, or
197 OH.
198
199 In addition, CRISM spectra that exhibit the doublet feature also exhibit an
200 absorption at ~1.9 µm due to the combination stretch+bend vibration mode of H2O.
201 A similar doublet absorption has also been observed in laboratory spectra of poorly
202 crystalline Fe-silicate and sulfate mixtures [Tosca et al., 2008b] as well as acid-
203 leached Fe-smectite [Madejova et al., 2009], suggesting the spectral signature may
204 also be consistent with poorly crystalline or amorphous phases. Therefore, while it
205 is likely that the spectral doublet feature is indicative of OH and/or H2O-bearing
206 phases, the specific mineral assignments remain ambiguous.
207
208 Roach et al. [2010] proposed four hypotheses to explain the doublet
209 material: (1) a mixture of sulfates (in particular, sulfates structurally similar to
210 gypsum and jarosite), (2) a mixture of opaline silica and Fe/Mg-smectite, (3) an Al-
211 OH phase mixed with Fe/Mg-smectite and/or jarosite, and (4) acid-leached Fe/Mg-
212 smectite or neoformation of poorly crystalline clays. Subsequent study by Weitz et
213 al. [2014] supported the latter two hypotheses based on an analysis of CRISM
214 spectra for the Melas Chasma region, demonstrating that the spectral doublet is
215 associated with “light-toned draping deposits” (termed LD in Weitz et al., 2014) that
216 locally form some of the highest stratigraphic units. These LD deposits extend
217 through portions of Melas Chasma and drape over pre-existing topography,
218 suggesting an airfall origin. Weitz et al. [2014] interpreted this LD unit to have been
10 219 deposited after the Late Hesperian valleys that incise the walls of the chasma, but
220 stratigraphic position alone cannot be used to determine whether the hydrated
221 phase(s) associated with the spectral doublet in this unit represent transported
222 sediments or authigenic products.
223
224 Roach et al. [2010] described a similar thin light-toned unit in Ius Chasma
225 that drapes pre-existing topography, but the stratigraphic and possible genetic
226 relationships between this unit and those in Melas have not yet been established.
227 Roach et al. [2010] interpreted the light-toned unit to have been emplaced on top of
228 Fe/Mg-smectite and sulfate-bearing strata either during or after a period of acidic
229 aqueous alteration when sulfates formed from evaporating brines. Another study by
230 Weitz et al. [2011], of Noctis Labrynthis, identified the doublet material in a ~20 m-
231 thick massive light-toned unit that outcrops near the top of a stratified light-toned
232 deposit [see Weitz et al., 2011 Figure 2]. It was suggested that this deposit was also
233 formed under acidic conditions, and the doublet-bearing unit was interpreted to
234 pre-date deposition of Fe/Mg- smectite and sulfates in the same region [Weitz et al.,
235 2011]. Although the doublet-bearing materials in Valles Marineris are often
236 associated with units that drape pre-existing topography, including the canyon
237 wallrock, Metz et al. [2010] provided evidence for deformation features in some of
238 the deposits, suggesting they may have originated higher up the canyon walls than
239 indicated by their current topographic position. Finally, CRISM data for Mawrth
240 Vallis also show evidence for doublet-bearing materials in the vicinity of Fe/Mg-
241 smectites and Al-phyllosilicates [Noe Dobrea et al., 2011; Bishop et al., 2013].
11 242
243 To summarize, doublet-bearing units are present in stratigraphic sections
244 that contain hydrated minerals suggestive of complicated and extended periods of
245 water-rock interaction and/or sediment transport. Fe/Mg-smectites and sulfates are
246 often found in nearby units and in most cases these occur in strata emplaced prior
247 to the doublet-bearing units. An exception is Noctis Labrynthis, where the doublet
248 material occurs in a unit that lies stratigraphically below an Fe/Mg smectite-bearing
249 outcrop.
250
251 We note that jarosite, which typically forms under conditions of pH<4, is also
252 identified in several of the regions that also exhibit the spectral doublet [Milliken et
253 al., 2008, Metz et al., 2009, Noe Dobrea et al., 2011]. Although the jarosite deposits
254 are not necessarily in rocks that are stratigraphically adjacent to the doublet
255 materials, the presence of this mineral in these regions may provide insight into
256 local aqueous geochemistry. Jarosite deposits on Mars may have formed where
257 acidic sulfate-bearing water ponded and subsequently evaporated [Tosca and
258 McLennan, 2006], though it is also possible for jarosite to form through chemical
259 weathering of basalt [e.g., Madden et al., 2004]. Alternatively, acidic conditions that
260 lead to jarosite formation may be induced by local or regional oxidation of Fe2+-
261 bearing fluids, possibly via the emergence of groundwater and interaction with the
262 oxidizing martian atmosphere [Hurowitz et al., 2009]. An important aspect of any of
263 these scenarios is that the presence of authigenic jarosite records information about
264 local pH as well as redox conditions.
12 265
266
267 2.3 Overview of Field Region at Rio Tinto, Spain
268 The headwaters of the Rio Tinto near Nerva, Spain are acidic (mean pH of
269 2.3) and rich in ferric iron and other heavy metals. Three terraces of the Rio Tinto
270 are exposed in the river valley with the youngest terrace closest in elevation to the
271 modern day channel and the intermediate and oldest terraces located
272 stratigraphically above it (see Figure 2). The youngest of these (called the lower or
273 young terrace) is Holocene in age (750-1500 years) and composed of poorly sorted
274 conglomerates overlain by cross-bedded sandstones. Both types of deposits are
275 consistent with formation by mass wasting and sediment accumulation in the active
276 channel, a process that also occurs in the modern river. The Pleistocene-aged (~35
277 Ky) intermediate terrace is exposed 100 m above the lower terrace and consists of
278 poorly sorted clasts of country rock within a goethite-bearing matrix; this terrace is
279 distinct from the other river terraces as it lacks the sedimentary facies
280 corresponding to channel and backwater deposits seen in both the young and old
281 terraces. Finally, the Late Pliocene-aged (2 My) upper or old terrace is exposed in a
282 ≥10m thick outcrop at Alto de la Mesa. The outcrop at Alto de la Mesa consists of a
283 lag conglomerate overlain by sandstones and iron oxide precipitates, reflecting
284 similar types of channel deposits as those seen in the lower terrace. Further
285 description of the terraces and radiometric age constraints are reported in
286 Fernández-Remolar and Knoll [2008].
287
13 288 The mineralogy of the modern river system and its terraces is dominated by
289 ferric oxides and ferric sulfates. A number of previous studies have investigated
290 variations in the mineralogical assemblages in the river terraces using X-ray
291 diffraction (XRD), visible-near infrared (VIS-NIR) reflectance spectroscopy, and
292 Raman spectroscopy [Fernández-Remolar et al., 2005; Roach et al., 2006; Chemtob
293 et al., 2006; Sobron et al., 2009, 2014,]. The reader is directed to Fernández-Remolar
294 et al. [2005] for a more thorough discussion on the mineralogy and mineral-forming
295 processes of the Rio Tinto. Those authors found that seasonal precipitation of Ca-
296 and Fe-sulfates occurs near the source region of the river, with the Fe-sulfates
297 jarosite (in particular hydronium jarosite), schwertmannite, and copiapite making
298 up much of the mineral assemblage, along with nanophase iron oxide and
299 nanophase goethite. Only the most stable minerals (iron oxides, goethite, jarosite)
300 survive through the wet season, during which meteoric water dissolves many of the
301 precipitates and aids in cementation of the terrace deposits. Jarosite is less
302 abundant and often absent in the older deposits, likely due to its long-term
303 instability under wet conditions, and these units instead contain more stable Fe-
304 oxides such as goethite and hematite. Therefore, the persistence of jarosite may
305 indicate a low degree of water-rock interaction or low water-to-rock ratios since its
306 precipitation. Country rock in this region is composed of quartz, chlorite, and pyrite,
307 which act as a source of solutes for the precipitates as well as a source of detritus for
308 the terrace conglomerates.
309
14 310 In addition to sulfates and Fe-oxides, phyllosilicates (2M1 illite, chlorite,
311 kaolinite, and smectites) are also present throughout the Rio Tinto system and are
312 thought to form by alteration of the volcanosedimentary bedrock near the source of
313 the acidic river system [see Fernández-Remolar et al., 2011 for details]. At Rio Tinto,
314 phyllosilicates typically accumulate at times of maximum waterflow during the wet
315 season when deposition of detrital sediment is dominant over chemical
316 sedimentation, but they can also be found in environments where conditions are
317 outside those typical for clay formation, such as the low-pH source streams. The
318 diversity and stability of these phyllosilicates depends on the hydrology, source rock
319 composition and geochemical processes occurring in the river, all of which must be
320 accounted for when interpreting the deposits [Fernández-Remolar et al., 2011].
321 Likewise, jarosite is neither stable nor formed in all of Rio Tinto’s hydrogeochemical
322 environments [Fernández-Remolar et al., 2005 and discussed below].
323
324 3. Methods
325 3.1 Hyperspectral Imaging Data
326 3.1.1 HyMap Data Analysis
327
328 For remote observations of Rio Tinto we used aerial HyMap scenes that
329 covered the headwaters of the river, nearby mining activity, and urban areas. These
330 data were acquired by Integrated Spectronics on Aug 14, 2004 and are the same as
331 the fully processed and analyzed data used by Roach et al. [2006]. The reader is
332 directed to that publication for a detailed description of how these data were
15 333 calibrated, atmospherically corrected, and processed. Briefly, HyMap is an aerial,
334 hyperspectral system with 126 channels covering a wavelength range of 0.45 to 2.5
335 μm at a spatial resolution of ~8m per pixel. Spectra from the processed images were
336 examined for individual pixels as well as spectral averages from multiple pixels that
337 correspond to larger regions of geologic or spectral interest, called regions of
338 interest (ROIs). Wavelengths from ~1.3 - 1.45 μm and ~1.8 - 2 μm were excluded
339 due to atmospheric absorptions. A principal component analysis (PCA) was carried
340 out for the data to reduce spectral variance and to highlight regions with similar
341 spectral properties (Figure 3). PCA is a purely mathematical procedure and does not
342 rely on a user’s a priori knowledge of variability in a region; thus, it can be difficult
343 to attribute different features observed in PCA images to unique differences in
344 mineralogy. However, as a purely qualitative tool we found this method reliable for
345 highlighting regions in the HyMap scenes whose spectra exhibited the doublet
346 absorption feature (e.g., Figure 3).
347
348 3.1.2 CRISM Data Analysis
349
350 CRISM is a hyperspectral imaging system on NASA’s Mars Reconnaissance
351 Orbiter that consists of several spectrometers. The “L” spectrometer covers near-
352 infrared wavelengths from 1.0 to 3.9 μm using 431 channels, has a spectral
353 resolution of ~6.55nm per channel, and a maximum spatial resolution of ~15-19 m
354 per pixel [Murchie et al., 2007]. The highest spatial-spectral resolution images are
355 Full Resolution Targeted (FRT) observations, and for this study we analyzed three of
16 356 these images in detail from the western Valles Marineris region near Melas and Ius
357 Chasma (see Figure 4, Figure 9). A full description of CRISM image processing
358 methods is provided in the supplemental material.
359
360 The variability in the doublet absorption feature was studied in detail by
361 handpicking spectra from geologic units and ROIs that exhibited these spectral
362 features. The resulting spectra commonly represent 3x3 or 5x5 pixel averages.
363 Each spectrum was fit with an upper convex hull to define the spectral continuum
364 over the ~2.1 to 2.35 μm wavelength range. The original spectra were then divided
365 by these continuum fits to produce continuum-removed spectra. This step isolates
366 the wavelength region where the doublet occurs and is necessary to compare band
367 depths between spectra [Clark and Roush, 1984]. Band depth is defined as 1-[Rb(λ
368 )/Rc(λ)], where Rb represents the observed reflectance value at wavelength λ and
369 Rc represents the calculated reflectance of the spectral continuum fit at the same
370 wavelength. Band depth maps were created for each CRISM image to provide a more
371 in-depth analysis of the spatial distribution of the spectral doublet feature.
372
373 Gaussian curves were also fit to the continuum-removed doublet absorptions
374 and the resulting Gaussian amplitudes and center positions were mapped for each
375 CRISM image. Gaussian modeling of absorptions is commonly used for remote
376 sensing data to assess band strength in the presence of noise or to remove the
377 influence of adjacent or partially overlapping absorption features on the absorption
378 of interest [Clark and Roush, 1984; Sunshine and Pieters, 1993]. Noise in the CRISM
17 379 images, for instance, may introduce apparent changes in band position/strength if
380 these calculations are based on the observed absolute reflectance values on a pixel-
381 by-pixel basis. This can introduce uncertainties in the band depth maps, which we
382 attempt to mitigate by examining the amplitudes and positions of Gaussian fits.
383
384 To determine the background noise in an image prior to Gaussian fitting, a
385 histogram of the standard deviations within each image was examined for a flat,
386 absorption-free portion of the spectrum (1.70-1.85 μm), an example of which is
387 presented in Figure 4. Here it is clear that the standard deviations in the ratio
388 spectra are predominantly <0.005, thus any absorption feature (band depth) in the
389 ratio spectra that is larger than this value is above the typical level of ‘noise’ in the
390 CRISM image. Two CRISM ratio spectra exhibiting the doublet feature and their
391 corresponding Gaussian fits are also presented in Figure 4, where it is clear that the
392 doublet absorptions are much stronger than the typical channel-to-channel
393 variation in reflectance. Pixels that exceed the 0.005 threshold are fit with two
394 Gaussian curves, one for each absorption in the doublet (see example in Figure 4).
395
396 3.2 Field Measurements
397 Fieldwork was conducted at multiple locations in the vicinity of Rio Tinto,
398 Spain in early May of 2014. Rock and/or soil specimens were collected and field
399 spectra were acquired at each of five primary sites (Table 1). These locations
400 included the modern streambed environment and two successively older terraces
401 that have been affected to different degrees by diagenetic processes (Figure 2).
18 402 Previous studies have shown that the materials at these sites exhibit different
403 mineral assemblages [Fernández-Remolar et al., 2005; Fernández-Remolar and
404 Knoll, 2008], allowing us to examine whether spectral reflectance properties can be
405 linked to post-depositional processes in these types of environments.
406
407 Samples collected were collected from the Rio Tinto source region (Anabel’s
408 Garden, Richi Spring, and Source Spring sites), a series of small streams with
409 actively forming efflorescent and crust-forming precipitates (e.g. copiapite, jarosite,
410 gypsum). A number of samples were also collected from an actively mined terrace
411 directly above Anabel’s Garden that showed evidence for the spectral doublet
412 absorption in the airborne HyMap data. Another field site, Berrocal, is also located
413 directly on the Rio Tinto approximately 20 km downriver from the mining
414 operations. At this location, there is a mix of detrital sediment and authigenic
415 precipitates. We also examined older deposits at Barranco de los Locos, a site that is
416 part of the Holocene-aged lower terrace in which the lower unit is a ferruginous
417 conglomerate and is overlain by sandstone [Fernández-Remolar et al., 2005].
418 Finally, for comparison to the modern environment we examined an older (early
419 Quaternary) terrace exposed at Alto de la Mesa. Detailed descriptions of these sites,
420 their ages, and local geology can be found in previous publications [e.g., Fernández-
421 Remolar et al., 2005; Fernández-Remolar and Knoll, 2008; and references therein].
422
423 A portable VIS-NIR spectrometer (ASD FieldSpec 3®) was used to acquire
424 reflectance spectra in situ from 0.35 to 2.5 μm. Reflectance spectra were acquired
19 425 for a variety of materials passively (relying on ambient sunlight) and/or actively
426 (relying on a contact probe with internal illumination). All measurements were
427 made using a Spectralon® panel as the reflectance standard. Passive measurements
428 that rely on solar illumination and reflectance are more directly analogous to
429 airborne data such as HyMap. A more detailed description of the ASD measurements
430 is provided in the supplemental material.
431
432 Rock and sediment samples were collected to capture the spectral diversity
433 of deposits at each location and were based in part on interesting absorption
434 features observed in the HyMap and field reflectance spectra (see Table 1 for full
435 list). To the extent possible, samples were collected at the different sites to be
436 representative of the mineralogy expected at HyMap and CRISM spatial scales (e.g.,
437 the dominant mineral assemblages at scales of ~8 - 18 m/pixel, respectively),
438 focusing on regions whose spectra exhibited evidence for compositional diversity.
439 Of specific interest were regions with spectral or field-based evidence for Fe-oxides,
440 Fe-sulfates, and clay minerals. A number of locations contained materials whose
441 reflectance spectra exhibited the doublet absorption feature described above, and a
442 variety of these materials were collected for detailed laboratory characterization.
443 Though many other samples and spectra were collected during the field campaign,
444 the samples, field measurements, and HyMap data whose spectra exhibit these
445 distinct doublet absorptions were chosen as the focus for the present study. A more
446 complete treatment of how field-based spectral observations and diversity at Rio
20 447 Tinto relate to aerial and orbital-scale observations is presented in Roach et al.
448 [2006].
449
450 3.3 Laboratory Measurements
451 3.3.1 Sample Preparation
452 All laboratory analyses were conducted in the Department of Earth,
453 Environmental and Planetary Sciences at Brown University. A number of small chips
454 (~2.5 cm) were removed from each hand sample and a second portion of each was
455 ground by hand in an agate mortar and pestle. The resulting powders were ground
456 such that the material could pass through a <45 μm mesh sieve. Grinding and
457 sieving serves the dual purpose of homogenizing the sample and reducing particle
458 size-dependent absorption/scattering processes. Where possible, separate powders
459 were prepared for different portions of a single sample to assess heterogeneity (e.g.
460 rock varnish, dark or light components, areas with different degrees of weathering)
461 and for comparison to the whole rock powders. Some hydrated mineral phases are
462 expected to be unstable even over short time frames, and these are not included in
463 our dataset.
464
465 3.3.2 Reflectance Measurements
466
467 Reflectance spectra of the hand samples, chips, and powders were measured
468 in the laboratory using the same ASD FieldSpec 3 instrument over a wavelength
469 range of 0.35 to 2.5 μm. The measurement spot size was adjusted to encompass the
21 470 entire sample in order to provide a better assessment of the bulk spectral properties
471 of each hand, chip, and powder sample rather than averaging spectra of individual
472 components. We note that the chips and powder samples were placed in ~10 mm
473 diameter sample holders for measurement, thus the measurement spot size for
474 these samples is equivalent.
475
476 All of the rock powders sieved to <45 µm were also measured with a Nicolet
477 iS50 FTIR from 1.75 – 25 μm. The FTIR was equipped with a Praying Mantis diffuse
478 reflectance attachment from Harrick Scientific and diffuse gold from Labsphere was
479 used as a reflectance standard. These measurements used a KBr beamsplitter and
480 DTGS detector, and each final spectrum was an average of 50 scans. Absolute
481 reflectance values from the FTIR are highly dependent on any slight differences in
482 height between the sample and gold standard, thus all FTIR spectra were scaled to
483 match the ASD values at 1.7μm. These spliced spectra provide full VIS-NIR-mid-IR
484 wavelength coverage, and because all samples were measured under identical
485 conditions, the observed relative differences between samples are reliable.
486
487 All reflectance spectra (field and lab) were processed using the same
488 methods, and continuum-removed spectra and band depths were calculated using
489 the same procedures as described above for the CRISM data (i.e., an upper convex
490 hull method was used to define the continuum slope). We analyzed the continuum-
491 removed spectra using Matlab to find the local minima (which is equivalent to the
492 maximum band depth) within the ~2.1 – 2.35 µm wavelength region. If the
22 493 continuum-removed spectrum is determined to exhibit absorptions near ~2.2 μm
494 and ~2.26 μm, then this so-called doublet feature is further analyzed to determine
495 the total width, center position, and depth of the individual absorptions.
496
497 3.2.3 X-ray Diffraction Measurements
498
499 The <45μm sieved powders were measured with a Bruker D2 Phaser XRD
500 system (see supplemental material for details) to determine qualitative and
501 quantitative mineralogy. Samples whose corresponding reflectance spectra
502 exhibited the 2.2 μm doublet (n = 8) were studied in detail, as were a subset of
503 samples from multiple locations whose spectra did not exhibit the doublet (n = 3) in
504 order to test our hypothesis that a distinct mineral assemblage gives rise to the
505 doublet feature (Table 2). Powder diffraction patterns were typically acquired over
506 a range of 2-90° 2θ, 0.02˚ step size, with a total integration time of up to ~12 hours
507 (6.5 s/step).
508
509 XRD patterns were analyzed with two programs, DIFFRAC.SUITE™ EVA for
510 full pattern phase identification and DIFFRAC.SUITE™ TOPAS for quantitative
511 Rietveld refinement, where the latter is similar to the method described in Bish and
512 Howard (1988). An overview of the mineralogy at Rio Tinto is provided in
513 Fernández-Remolar et al. [2005], which we used as a guideline for choosing
514 reasonable starting components for the qualitative mineral assessment. Modal
515 mineralogy estimated from Rietveld refinement relied on the structural and
23 516 instrumental parameters to model the XRD patterns. Structure files describing the
517 crystallographic structure of relevant minerals used in the refinement are listed in
518 the supplementary table.
519
520 These mineral structures were added and removed from the model
521 iteratively until a reasonable fit to the observed pattern was obtained. TOPAS
522 automatically normalizes the model results such that mineral abundances will
523 always add to 100%. The quality of the refinement is determined by the weighted
524 profile R-factor (Rwp), which considers the root mean square error (RMSE) for each
525 scaled intensity value (peak and background). Although a smaller Rwp generally
526 indicates a better model, it is possible to over fit the data with this method and
527 discretion is used to ensure the resulting model is geologically realistic. We also
528 calculated the areas for the muscovite 001 (~9° 2θ) and jarosite 101 (~15° 2θ)
529 peaks. Though this method does not provide quantitative estimates like the Rietveld
530 method, it does provide an independent estimate of how the relative abundances of
531 the two minerals vary between samples without the uncertainties associated with
532 modeling the full XRD pattern.
533
534 4. Results
535 4.1 Overview of Rio Tinto Field Observations
536 Previous work has shown that mineralogical and spectral differences exist
537 between the old and young terraces [Fernández-Remolar et al., 2005; Sobron et al.,
538 2014], and we observe congruous spectral differences in our samples from these
24 539 locations. Field spectra of samples from the source region exhibit absorptions
540 consistent with the presence of gypsum, jarosite, and copiapite (Figure 5). Hydrated
541 iron-bearing phases are also common in these field spectra, including absorptions
542 consistent with the presence of schwertmannite and copiapite, which are identified
543 in spectra from the Richi Spring and Source Spring locations. The field spectra from
544 Berrocal (not shown) have very similar identifying absorptions that are typical of
545 the precipitates, phyllosilicates and iron hydroxides found in the source region
546 spectra. Spectra exhibiting the doublet absorption are found in sand bar sediments
547 from Berrocal, representing rocks/sediments that have not been influenced by
548 mining activities.
549
550 Most of the Fe- and Ca-sulfate precipitates are absent at the young terrace
551 site, Barranco de los Locos. Jarosite, Al-phyllosilicate (muscovite/illite), and goethite
552 are the primary minerals identified in our field spectra at this location. VIS-NIR
553 spectra from the old terrace site, Alto de la Mesa, exhibit the lowest mineral
554 diversity and are dominated by iron oxides. Most spectra from Alto de la Mesa are
555 consistent with the presence of goethite, though some indicate the presence of
556 kaolinite or chlorite. Sulfate absorptions are not apparent in spectra of materials
557 from the old terrace, and the doublet feature is also not observed in these spectra.
558 However, spectra for a number of samples that were taken from a mine road terrace
559 above Anabel’s garden display the doublet absorption. Other samples from this site
560 are spectrally consistent with goethite and hematite. Finally, samples from some of
25 561 the mining terraces are spectrally consistent with chlorite, kaolinite, and muscovite,
562 likely reflecting country rock components.
563
564 4.2 Laboratory Measurements
565 4.2.1 X-ray Diffraction
566
567 Based on an initial qualitative assessment of the XRD patterns and observed
568 spectral diversity, eleven samples were chosen for further quantitative Rietveld
569 refinement. Eight of the samples have spectra that exhibit the spectral doublet and
570 contain between 3 - 38% jarosite and 4 - 50% illite or muscovite (Table 2). The
571 other volumetrically important mineral in these samples is quartz, which does not
572 have absorptions at VIS-NIR wavelengths, so the samples are spectrally dominated
573 by Fe-sulfate and Al-bearing clay minerals. K-jarosite is a better fit in the Rietveld
574 refinement than either hydroniumjarosite or natrojarosite. Though the mineral
575 assemblage is largely the same, the relative proportion of jarosite and Al-clay is
576 quite variable between samples (Figure 6), and these values are presented in Table
577 2. Three samples that do not show evidence for the doublet feature in
578 corresponding reflectance spectra were also examined. The XRD patterns for these
579 samples are consistent with a mineral assemblage dominated by hematite, goethite,
580 kaolinite, chlorite and quartz, but no jarosite, in agreement with the findings of
581 Fernández-Remolar et al., [2005] for this locale (Table 2). One sample from the
582 young terrace, Barranco de los Locos sample RT23, is modeled to contain ~33%
583 jarosite and ~39% gypsum but no Al-clay (Table 2), and its reflectance spectrum is
26 584 dominated by gypsum at wavelengths >1.3 µm (Figure 5). In summary, the major
585 secondary phases in these samples are jarosite, illite or muscovite, and Fe-oxides
586 such as goethite and hematite. Although smectitic clays are an integral part of the
587 alteration mineral assemblages on Mars, the Rio Tinto samples examined for this
588 study show no clear evidence for the presence of smectitic clays.
589
590 4.2.2 Reflectance Spectra
591
592 As with the field spectra, the laboratory spectra indicate variations in
593 mineral assemblage with age. In samples from the modern streambeds (samples
594 from Anabel’s Garden and Source Spring) gypsum is the primary mineral identified
595 in two samples (RT18, RT23), with jarosite also being present in RT23 (see Figure
596 5). Other samples from these sites have spectra with strong OH and H2O bands at
597 ~1.4 and ~1.9 µm as well as absorptions consistent with copiapite, schwertmannite,
598 or possibly monohydrated sulfates (Figure 5). These hydrated samples were not
599 measured with XRD because these minerals change hydration states readily under
600 laboratory conditions, so these spectral mineral identifications cannot be confirmed.
601 However, other XRD studies of Rio Tinto samples have found evidence for a variety
602 of sulfates, including epsomite, copiapite, and coquimbite [Fernández-Remolar et al.,
603 2005], and it is likely that a range of hydrated sulfates are also present in our
604 samples.
605
27 606 The spectral shapes of samples from the young terrace are distinct from
607 those of Anabel’s Garden and Source Spring, with the former having weaker
608 hydration absorptions and a strong spectral slope near 1 μm. In fact, the major
609 difference in spectra of all the young terrace samples lie in the region between 2.2
610 and 2.3 μm; spectra of some samples show a doublet in this region whereas others
611 have only an ~2.21 or 2.27 μm absorption. Samples collected from the old terrace
612 deposit are spectrally consistent with clays such as kaolinite and chlorite as well as
613 Fe-oxides such as hematite (Figure 5). Gossan samples all show spectral evidence of
614 goethite. As with the field spectra, lab measurements of the older samples typically
615 lack spectral evidence for sulfates and are instead indicative of Fe-oxides and/or
616 phyllosilicates. Samples from the mining terrace have spectra that exhibit the
617 spectral doublet or that are representative of the same types of minerals discussed
618 for the other sites, including chlorite and Fe-oxides (goethite and hematite).
619 However, given that the specific geologic context of these samples is unknown,
620 especially compared to the well-defined mineralogical and age progression of the
621 river terraces, it is difficult to comment on the processes that have affected this
622 group of samples.
623
624 To summarize thus far, laboratory spectra with the doublet absorption
625 correspond to samples collected from the modern stream sites (Anabel’s Garden,
626 Source Spring), the young terrace (Barranco de los Locos), and from a mining
627 terrace site near Anabel’s Garden (Table 2). There are many minerals that have
628 spectral features near 2.26 µm, but we attribute the 2.26 µm feature in the doublet
28 629 to jarosite because of the combined presence of absorptions at wavelength positions
630 that are attributed to jarosite (1.48, 1.86, 1.9, 2.26 µm), as well as evidence from
631 XRD mineralogical analysis. The doublet feature is conspicuously absent in spectra
632 of samples collected at the old terrace, all of which lack jarosite. Reflectance spectra
633 of the <45 µm powders of samples exhibiting the doublet show strong, broad
634 electronic transition absorptions at wavelengths of ~1 μm due to the presence of
635 iron (Figure 6). There are also narrower vibrational absorption features at 1.4 and
636 1.48 μm due to OH, though the 1.48 μm feature is weak or absent in samples that
637 show a very strong 2.21 μm short wavelength band in the doublet feature. Likewise,
638 a broader absorption at ~1.9-1.95 μm due to OH/H2O and a narrower absorption at
639 1.86 μm from OH/H2O are also weaker in samples with a stronger ~2.21 µm
640 feature.. In contrast, the ~2.21 µm absorption in the doublet feature is related to the
641 presence of Al-phyllosilicates (illite or muscovite). A vibrational absorption near
642 2.35 μm is present in some samples and is also consistent with the presence of Al-
643 OH in illite or muscovite.
644
645 Spectra of rock chips and hand samples were also acquired and, in most
646 cases, the samples whose powder spectra exhibited the doublet feature also exhibit
647 the feature for these larger physical forms (see example for sample RT19 in Figure
648 7). However, sample RT10 is an exception in that a spectrum of the orange-colored
649 varnished face of the hand sample shows the doublet feature whereas the
650 unvarnished side of the hand sample, a chip, and the powder spectra are dominated
651 by the 2.21 μm feature (Figure 7). This indicates the varnish may be enriched in
29 652 jarosite, and although such coatings or varnishes may be only several micrometers
653 thick, they may contribute significantly to aerial and orbital spectral observations.
654
655 Only the bulk powder versions of the samples were measured using both
656 reflectance spectroscopy and XRD; thus, we compare only the powder spectra
657 directly with the mineralogy determined by XRD. There is a strong correlation
658 between the strength of the longer wavelength absorption in the doublet feature
659 and the abundance of jarosite for the samples examined here (Figure 8). There is
660 also a strong relationship between the relative band strengths and the relative
661 amounts of jarosite and Al-clay in the samples, as determined using either the
662 Rietveld modal estimates or the XRD peak area. The relationship between Al-clay
663 and the strength of the shorter wavelength absorption has a lower R2 value than in
664 the other plots, but we suggest this is due to the size of the dataset, where a
665 relatively small amount of scatter (in this case one data point with anomalously high
666 estimated Al-clay wt.%) has a strong control on the fit. These observations indicate
667 that the doublet feature observed in spectra of some samples from Rio Tinto is
668 related to the mixing of Al-phyllosilicate (illite or muscovite) with jarosite, and that
669 the strengths of the absorptions within the doublet feature are directly related to
670 the relative proportion of these phases.
671
672 4.3 HyMap and CRISM Observations
673 4.3.1 HyMap Results
674
30 675 The HyMap PCA false-color map presented in Figure 3 highlights pixels
676 whose spectra exhibit the doublet feature in dark red. Absorption maxima
677 (reflectance minima) for these pixels occur at ~2.20-2.22 and 2.26-2.28, a range that
678 is broader than observed in Mars CRISM data and that could be a result of
679 differences in spectral resolution, since HyMap only has 6 bands in this wavelength
680 region compared to CRISM’s 13 bands. The doublet feature is evident in HyMap
681 spectra corresponding to the Anabel’s Garden location, where it is possible to track
682 changes in the relative strength of the absorptions across a ~500 m transect in the
683 HyMap image (Figure 3). The feature is also present in numerous pixels
684 corresponding to mining terraces, although it was not possible to independently
685 examine most of these locations during the field study. Bedrock, detritus, and soil at
686 the young, intermediate, and old terrace sites discussed above are not always
687 apparent in the HyMap data due to the presence of vegetation. However, chlorite,
688 kaolinite, gypsum, and Al-clay are identifiable in the HyMap spectra, and Roach et al.
689 [2006] provide a thorough analysis of iron oxides and hydrated or hydroxylated
690 iron-sulfates in these airborne data. In this study we have only relied on the HyMap
691 data to provide an airborne comparison (at lower spatial and spectral resolution) of
692 the doublet-bearing materials with the spectra acquired in the lab and in situ using
693 the ASD FieldSpec3. In this context, the HyMap data confirm that spectra of certain
694 surface materials exhibit the doublet feature even at a meters-per-pixel spatial scale
695 and lower spectral resolution, consistent with the higher spatial and spectral
696 resolution field and laboratory data.
697
31 698 4.3.2 CRISM Results
699
700 The spectral doublet feature is evident in a number of CRISM images in the
701 Valles Marineris, with particularly good examples in Melas and Ius Chasma [e.g.
702 Roach et al., 2010, Weitz et al., 2014]. Our Gaussian fit analysis of three of these
703 CRISM images show that the absorption band centers and relative strengths
704 (Gaussian amplitudes) associated with the doublet materials are only moderately
705 variable (Figure 9). The centers of the Gaussian fits for the three CRISM images and
706 the HyMap image are listed in Table 3. The typical 2σ deviation on the band
707 positions in the CRISM images is ~0.015 µm and the band centers fall near 2.2060
708 and 2.2682 µm on average. However, the wavelength position varies across the
709 spatial dimension of the CRISM detector (spectral smile) by ~0.010 µm, which may
710 explain much of the width of the histograms in Figure 9. To examine this effect we
711 analyzed a 50 x 50 pixel region restricted to the center of each image where the
712 spectral smile is minimal. Histograms for these subsets reveal a decrease in
713 standard deviation as well as a shift in mean band center in all three cases (Table 3,
714 Figure 4). The difference in band centers between the full image and the central
715 subset is statistically significant (p < 0.05) in every case.
716
717 Therefore, we conclude that the band centers of the doublet spectral feature
718 are relatively constant and occur at wavelengths of ~2.21 and ~2.265-2.270 µm. It
719 should be noted that Weitz et al. [2014] identified another spectral doublet
720 occurrence with a long wavelength absorption closer to ~2.28 µm, and the inclusion
32 721 of pixels with this feature in our analysis would create additional spread in the
722 histograms in Figure 9c. Given that we did not analyze each pixel in each image
723 individually, it is possible that this longer wavelength doublet is present, but it is
724 certainly not a dominant spectral signature, which is evident in the histograms and
725 investigation of spectra from individual pixels and ROIs.
726
727 While the wavelength positions of the two absorptions appear to be fixed,
728 their relative strengths are variable and are not directly correlated (Figure 9d).
729 Though this result is qualitatively similar to what is observed in field, lab and
730 HyMap spectra of materials from Rio Tinto, the range in relative amplitudes of the
731 two absorptions is typically much smaller in the CRISM data than for the terrestrial
732 examples. That is, spectra of some materials from Rio Tinto are clearly dominated
733 by the ~2.21 or the ~2.27 µm feature, with the other feature being very weak or
734 absent. In contrast, nearly all CRISM spectra for the regions examined here exhibit
735 both features and in many instances both features are of equal strength and, while
736 variability does exist, it is less intense than the variability in strengths observed in
737 the Rio Tinto spectra.
738
739 The relative band strengths, which is the Gaussian amplitude of the 2.21 µm
740 absorption divided by the amplitude of the 2.27 µm absorption, for the three CRISM
741 images is 1.1925±1.47 (2σ), 1.2714±1.35 (2σ), and 1.2207±1.28 (2σ). These
742 averages suggest that the band strength of the shorter wavelength absorption is
743 often slightly stronger than the longer wavelength absorption. However, the
33 744 standard deviation of the distribution is large enough that it is clear the two
745 absorptions do not co-vary. That is, the two absorptions that comprise the doublet
746 feature in CRISM spectra are independent of one another and thus represent two or
747 more distinct phases. Both tails of the distributions in Figure 9d are affected by
748 “bad” pixel values (typically a spike in the value for a single CRISM channel in one
749 column of the image that will influence the Gaussian fits). This is particularly
750 obvious as a bulge in the distribution as the Band 1/Band 2 amplitude ratio
751 approaches 0 (due to Band 1 amplitudes approaching 0) as well as by the long tail as
752 the Band 2 amplitude approaches 0, starting at a band amplitude ratio of ~3.
753 However, other than the extreme tails of the distributions (which represent a
754 relatively small number of pixels), these pixels play a minimal role in the overall
755 distributions of the band amplitude ratios.
756
757 4.4 A Case Study: Doublet-Bearing Materials in Ius Chasma, Mars
758
759 This study is focused on the spectral characteristics of materials found on
760 Mars and how they relate to those in Rio Tinto as seen in field, airborne, and lab
761 data, with the goal of assessing whether or not the latter may be appropriate
762 mineralogical analogs. However, to fully understand the processes and
763 environmental conditions associated with the deposits on Mars, and whether or not
764 Rio Tinto is an appropriate process analog, it will be necessary to conduct detailed
765 mapping studies that place the martian examples in their proper geologic context.
766 We conducted a small-scale examination of MRO HiRISE and CTX images of the
34 767 doublet-bearing materials in Valles Marineris to get a sense of the stratigraphic
768 position of these deposits with respect to surrounding strata.
769
770 As discussed by Roach et al. [2010], some occurrences of the materials with
771 the doublet absorption are in proximity to Fe/Mg clay-bearing strata. However, in
772 many cases the contacts between units are obscured and bedding is not apparent,
773 making it difficult to determine the relative stratigraphic position of these units.
774 Despite these complications, it is clear that the doublet absorptions are commonly
775 associated with light-toned units that drape pre-existing topography and span a
776 range in elevation (e.g., Figure 10a) or as deposits within the topographically lower
777 portions of the chasma (e.g., Figure 10b). In some locations the doublet unit
778 surrounds the clay-bearing units, whereas in others the clay-bearing unit is located
779 higher up the canyon wall. The doublet material also occurs as matrix in
780 megabreccia deposits [Roach et al., 2010], an example of which can be seen in
781 CRISM image FRT0000A396 in Figure 9. We now discuss a case study for one region
782 in Ius Chasma where several doublet-bearing units are in proximity to Mg/Fe clay-
783 bearing strata, focusing on the relative stratigraphic position of these units in order
784 to assess the relationship between the observed mineral assemblages in a more
785 detailed geologic context.
786
787 The relative strength of absorptions in the doublet feature were mapped for
788 several CRISM observations in Ius Chasma and overlain on HiRISE and CTX images
789 (Figure 11). This map reveals both the full extent of the doublet-bearing units and
35 790 allows for further discrimination of spectral properties within and between the
791 different units. Each colored pixel meets the criteria of having band depth > 0.005 at
792 both 2.205 and 2.269 µm. Where the ratio of band depths (BD 2.205/2.269 µm) is
793 small, we interpret the unit to be enriched in the ~2.27 µm phase, and a higher ratio
794 represents the opposite, in a strictly relative sense. In some locations (Figure 11c)
795 there is a full transition between these cases within a single unit. In this example,
796 the unit is heavily brecciated and lacks clear stratification. This example fits within
797 a larger regional trend where the doublet-bearing units with the strongest ~2.27
798 µm absorption are in the south, and seemingly higher up the walls of the chasm,
799 which then transitions to a topographically lower region where the doublet has an
800 absorption with two bands of equal strength, and finally to the north where the
801 ~2.21 µm absorption becomes stronger in the floor units (see the red-to-green-to-
802 blue northerly progression in Figure 11b). In this example it is clear that the
803 doublet-bearing unit is quite heterogeneous with regard to the strength of the two
804 absorptions, suggesting the relative abundances of the phases responsible for these
805 absorptions are also highly variable within this single geologic unit.
806
807 A second example reveals changes in composition that are consistent with
808 geomorphologic changes and that conform to geologic contacts. Three doublet-
809 bearing units (“bright”, “light”, and “moderate”-toned) and a darker, stratified,
810 Mg/Fe clay-rich unit are shown in Figure 11d, with close-up HiRISE views of the
811 contacts between these units presented in Figure 12. The Mg/Fe clay unit has tilted
812 strata (dipping approximately north/northwest, see Figure 12b) and spectra for
36 813 these units do not meet the criteria for being “doublet bearing” (with the exception
814 of a few pixels that have high band ratios and therefore higher amounts of the ~2.21
815 µm component). These stratified Mg/Fe clay units are superposed by a light-toned
816 doublet-bearing unit. The significant difference in bedding geometry and the sharp
817 nature of the contact suggests there is an angular unconformity between these units,
818 with the clay strata predating the doublet-bearing unit (Figure 12b). This light-
819 toned doublet unit has a relative band strength ratio of ~1, meaning the absorptions
820 within the doublet are of equal strength (however, this does not necessarily imply
821 that there must be equal amounts of each component). Overlying this unit is a
822 second doublet-bearing unit that is moderate in tone but that exhibits similar
823 morphological characteristics as the light-toned unit (Figure 12c). At full HiRISE
824 resolution it is apparent that both the light and moderate-toned units are stratified.
825 The moderate-toned unit has a band strength ratio between 1.2 and 1.8, indicating
826 the shorter wavelength (~2.21 µm) absorption is stronger.
827
828 The third unit in this region corresponds to pixels with band ratios <0.9,
829 consistent with it being enriched in the longer wavelength (~2.27 µm) component
830 relative to the other doublet-bearing units in this location. This unit is the brightest
831 in grayscale HiRISE and CTX images and is overlain by the light-toned unit (Figure
832 12d). In this location there is no direct contact between the dark-toned Mg/Fe clay
833 strata and this bright unit, thus the relative stratigraphic order cannot be
834 determined. However, in nearby locations it appears that this bright-toned unit that
835 is characterized by a stronger ~2.27 µm band is also stratigraphically above the
37 836 Mg/Fe clay-bearing unit, though these individual outcrops are too small to identify
837 compositionally with CRISM. Therefore, the apparent increasing stratigraphic order
838 is: Mg/Fe clay strata – (angular unconformity) – bright unit – light unit – moderate
839 unit. The associated mineralogy above the unconformity appears to decrease in the
840 longer wavelength (~2.27 µm) component and increase in the shorter wavelength
841 (~2.21 µm) component upsection, at least in a relative sense.
842
843 Together, these two examples indicate that the relationships between the
844 minerals responsible for the doublet feature and the local stratigraphy are diverse.
845 In the first example there appears to be spectral variation, and thus relative changes
846 in mineral abundance, within a single geologic unit. In the second case the dominant
847 spectral variations appear to conform to stratigraphic boundaries, indicating
848 differences in relative mineral abundance as a function of geologic unit and
849 stratigraphic position.
850
851
852 5. Discussion
853 5.1 Mineralogy
854
855 The CRISM spectral doublet feature is similar to features observed in
856 airborne, field, and laboratory spectra of materials in and around Rio Tinto, Spain.
857 As discussed above, XRD analysis of select Rio Tinto samples demonstrates that this
858 spectral doublet arises from a mixture of Al-bearing phyllosilicates and jarosite. By
38 859 analogy, CRISM spectra exhibiting the doublet feature may also arise from a mixture
860 of Al-phyllosilicates and jarosite, one of the possibilities discussed by Roach et al.
861 [2010] and Weitz et al. [2014]. This interpretation is supported because (1) the two
862 absorptions that comprise the doublet feature vary independently, suggesting at
863 least two phases are present, (2) the absorption features within the martian doublet
864 examples occur at the same wavelength positions as observed in the Rio Tinto
865 spectra and are consistent with Al-OH and Fe-OH vibrations, and (3) the doublet-
866 bearing material occurs on Mars in the vicinity of other phyllosilicate and/or
867 jarosite-bearing deposits. Therefore, we conclude that Al-phyllosilicate-jarosite
868 mixtures in the Rio Tinto may serve as spectral analogs for the doublet materials
869 found in the Valles Marineris and elsewhere on Mars. However, whether or not the
870 Rio Tinto deposits may also serve as process analogs warrants further study, as
871 discussed below.
872
873 Other suggestions for the doublet feature have included opaline silica and
874 clay, mixtures of sulfates (e.g. gypsum and jarosite), and poorly crystalline Fe-SiO4-
875 bearing phases. Reflectance spectra of opaline silica do exhibit a broad absorption
876 feature from 2.2-2.26 μm [Milliken et al., 2008], but it has been noted that this
877 feature is typically too broad to match the doublet even when mixed with Fe-rich
878 clay such as nontronite [Weitz et al., 2014]. One of our Rio Tinto hand samples,
879 RT23, provides a natural example of an intimate mixture of jarosite and gypsum in
880 which there are similar proportions of both phases and no phyllosilicates (Figure 5).
881 The reflectance spectrum of this sample is clearly dominated by the H2O absorptions
39 882 characteristic of gypsum and an absorption near 2.2-2.3 μm that is wider than
883 observed for the typical doublet feature. In addition, although gypsum exhibits very
884 narrow absorption features near ~2.22 and ~2.27 µm, these are rather weak and
885 they are superposed on the much broader and stronger feature that spans the
886 ~2.08-2.3 µm region (see gypsum spectrum in Figure 1). By contrast, the
887 absorptions in the doublet feature do not appear to be superposed on any broader
888 feature and each appears to be much stronger and more clearly defined than the
889 narrow features observed at similar wavelengths in gypsum. In summary, it does
890 not seem possible to recreate the shape of the doublet feature through a
891 combination of only gypsum and jarosite, thus we consider the doublet feature to be
892 most consistent with a mixture of an Al-phyllosilicate and jarosite.
893
894 In our Rio Tinto samples the Al-phyllosilicate is dominated by illite or
895 muscovite and is likely to be detrital, sourced from the surrounding country rock. In
896 contrast, the jarosite, which in some samples occurs as a coating, is authigenic and a
897 natural product of the Fe- and S-rich, oxidizing, acidic conditions that characterize
898 the Rio Tinto system. The decrease in jarosite with progression in age suggests it is
899 not stable over geologic time in the continued presence of water, particularly for
900 circum-neutral and higher pH fluids. Indeed, we observe the doublet feature only in
901 spectra that contain jarosite, which are primarily associated with the youngest
902 deposits such as those at Anabel’s Garden and at the Source Spring. It is notable that
903 in our Rio Tinto samples the spectral doublet feature can be directly attributed to
904 relatively well-crystalline phases as there is no clear spectral or XRD evidence for
40 905 hydroxylated poorly crystalline phases that may result from the acidic leaching of
906 clay minerals. This is not to say that such processes do not occur in these locations,
907 but rather to note that the presence of acid-leached clays or hydrated poorly
908 crystalline precipitates are not required to explain the presence of the spectral
909 doublet feature in the samples examined here.
910
911 The same may not be true of the doublet-bearing materials on Mars. There,
912 acidic leaching of basalt and/or pre-existing smectitic clays may be an important
913 process for forming alteration minerals, including some sulfates such as jarosite
914 [e.g., Tosca et al., 2004; Golden et al., 2005]. As noted above, acid-leaching of clays
915 has also been invoked to explain the presence of the spectral doublet feature on
916 Mars [Tosca et al., 2008b; Roach et al., 2010; Noe Dobrea et al., 2011; Weitz et al.,
917 2014], and the occurrence of jarosite in some locations is indicative of low-pH
918 conditions, at least at a local scale. It is possible to make a poorly crystalline Fe-SiO4-
919 bearing material through the acid leaching of nontronite, and reflectance spectra of
920 such products can exhibit absorptions with band centers similar to those observed
921 in the doublet feature [Madejova et al., 2009; Roach et al., 2010]. In this scenario, the
922 relative strengths of the absorptions may vary based on the degree of acid
923 alteration. However, spectra of residues from the acid leaching of nontronite may be
924 more consistent with Si-rich, opaline-like materials, and in these cases the full width
925 of the resulting doublet feature is slightly wider than is observed in the martian
926 examples [Tosca et al., 2008b; Altheide et al., 2010].
927
41 928 As a different example, reflectance spectra of Fe/Mg smectite-rich altered
929 basalts from Mauna Kea, Hawaii that have undergone acid leaching exhibit a doublet
930 feature that is thought to arise from the presence of jarosite and hallyosite [Graff et
931 al., 2012]. In this instance the spectral doublet feature is the result of two distinct
932 (presumably authigenic) mineral phases likely produced by
933 dissolution/transformation of pre-existing phases and not associated with a poorly-
934 crystalline leached clay residue. The spectral characteristics of halloysite-jarosite
935 mixtures, as presented in Graff et al. [2012], are more consistent with the doublet
936 characteristics observed on Mars than are the spectral characteristics of a Si-rich
937 leached clay residue. Spectra of the latter tend to exhibit absorptions in the ~2.2-2.3
938 µm region that span a slightly broader wavelength range than the martian examples
939 and for which the individual absorptions are not as narrow or well-defined as the
940 martian examples. Halloysite and kaolinite are spectrally similar in the ~2.2-2.3 µm
941 region, thus halloysite-jarosite, kaolinite-jarosite, and montmorillonite/illite-
942 jarosite mixtures could all be consistent with the martian doublet materials.
943
944 Perhaps akin to the observations by Graff et al. [2012], laboratory
945 experiments by Tosca et al. [2008b] indicate that dissolution of basalt and addition
946 of sulfur could lead to fluids that form precipitates under oxidizing conditions
947 whose spectra also exhibit a doublet between 2.2 and 2.3 μm [see example spectrum
948 in Roach et al., 2010]. Though some of the precipitates generated during these
949 experiments were X-ray amorphous, others exhibited XRD patterns consistent with
950 neoformation of jarosite, Fe-oxides, and opaline silica [Tosca et al., 2008b]. VIS-NIR
42 951 reflectance spectra of these precipitates were also consistent with the presence of
952 opaline silica, minor jarosite, and an Al-OH bearing phase, though the mineralogical
953 host(s) of observed Al-OH bands at ~2.21 µm was not directly evident in the XRD
954 data and it/they may be below the XRD detection limit.
955
956
957 5.2 Formation Processes: Detrital and Authigenic
958
959 From a process point of view the distinction between the scenarios described
960 above may be somewhat semantic. The important point is simply the recognition
961 that dissolution and alteration of basaltic materials on Mars, including cases where
962 clay minerals may already be present, can lead to fluids capable of forming jarosite
963 and Al-rich phyllosilicates. However, co-occurrence in the martian deposits does not
964 necessarily imply that both jarosite and Al-phyllosilicates were formed at the same
965 time, under the same conditions, or even in the same region. Either mineral could be
966 detrital or authigenic in the doublet-bearing units, and detailed geologic context in
967 conjunction with an assessment of the full mineral assemblage is needed to assess
968 which scenario is most likely.
969
970 Given that halloysite, kaolinite, muscovite/illite, and montmorillonite all
971 exhibit an absorption feature at ~2.21 µm and that all have been identified on Mars
972 from CRISM data [Ehlmann et al., 2009], it may be difficult to unambiguously
973 determine which Al-bearing phyllosilicate is present in these purported mixtures
43 974 using VIS-NIR reflectance spectroscopy alone. There is evidence in some of the
975 martian spectra that the doublet feature may exhibit a shoulder on the short
976 wavelength side of the ~2.21 µm absorption, indicating a weak and narrow
977 absorption may be present in the ~2.15-2.21 µm region. If true, this could be more
978 consistent with the presence of kaolinite or halloysite as the Al-bearing
979 phyllosilicate as opposed to illite or muscovite or montmorillonite. The presence of
980 jarosite would indicate low-pH conditions (at least locally), and previous studies
981 have noted that kaolin minerals can be formed under acidic conditions [e.g., Fialips
982 et al., 2000]. On Mars, oxidation of fluids enriched in ferrous iron could lead to local
983 (and transient) acidification, which could promote jarosite formation if sulfides are
984 present or if sulfate is present in solution [e.g., Hurowitz et al., 2009, 2010].
985
986 The processes described above indicate the possibility that both the Al-
987 phyllosilicates and jarosite are authigenic and thus indicative of local aqueous
988 geochemical conditions. Though some Al-phyllosilicates can form under moderate
989 to low-pH conditions, particularly kaolinite, an alternative possibility is that the Al-
990 phyllosilicates in the martian doublet materials are detrital. In this scenario the
991 jarosite would be indicative of local aqueous conditions and the phyllosilicates
992 would reflect eolian or aqueous sediment transport processes. If true, the doublet-
993 bearing units in Valles Marineris may be similar to Rio Tinto in terms of mineralogy
994 and process, in which case the relative proportions of Al-phyllosilicate and jarosite
995 in a given deposit could reflect the relative contributions of sediment input and
996 mineral precipitation, respectively. Recent experimental work by Dixon et al. [2015]
44 997 demonstrated that jarosite dissolution is not strongly dependent on hydrologic
998 conditions and that 1 mm jarosite particles are unlikely to survive for more than 104
999 - 105 years even at low flow rates (<0.01 L hr-1). Therefore, while it is possible that
1000 the jarosite on Mars may be detrital, this is unlikely to be the case for long-lived
1001 fluvial transport systems, similar to the lack of jarosite observed in the older
1002 deposits at Rio Tinto. That stated, because we cannot constrain the particle size of
1003 the jarosite we cannot rule out eolian transport for this component.
1004
1005 A detailed ‘source to sink’ and stratigraphic study of the geologic setting of
1006 each of the doublet bearing units on Mars is needed to help determine which ones
1007 may represent diagenetic/authigenic phases (do observed mineral boundaries cut
1008 across the stratigraphy?) versus those that may be detrital (do variations in
1009 mineralogy conform to stratigraphic boundaries?), which is beyond the scope of the
1010 current study. However, given the diversity in morphologic properties of the
1011 doublet-bearing units within Valles Marineris and the results of the case study
1012 described above, it is likely that multiple formation processes are responsible even
1013 though the spectral signatures are similar.
1014
1015 The massive brecciated deposit in Ius Chasma that shows a relative
1016 transition from more jarosite to more Al-phyllosilicate within a single geologic unit
1017 (first example in our case study, Figure 11c) could be evidence of in situ alteration.
1018 Similarly, the thin draping units such as the one shown in Figure 10a are difficult to
1019 explain by processes other than airfall deposition [e.g., Weitz et al., 2015], which
45 1020 would suggest a relatively fine grain size and thus make the original sediment more
1021 susceptible to chemical alteration if it has not been welded. In this case, the draping
1022 unit may be an airfall dust or volcanic ash deposit, either of which could contain
1023 sulfur, that was later altered by water, possibly in the form of thins films from
1024 melting snow or ice. Atmospheric sources of sulfur (e.g., from volcanic emissions)
1025 could also contribute to authigenic sulfate formation. Regardless of the source, local
1026 differences in iron and/or sulfur content of the original sediment would lead to
1027 different authigenic mineral assemblages, such as different amounts of Al-
1028 phyllosilicate and jarosite [Michalski et al., 2013].
1029
1030 The second example in our case study of Ius Chasma , where Mg/Fe clay-
1031 bearing strata are unconformably overlain by layers have differing amounts of
1032 inferred jarosite and Al-phyllosilicate (Figures 11-12), may be more consistent with
1033 changes in sediment flux and/or source region. The stratigraphic order of these
1034 units indicates the large Fe/Mg clay unit pre-dates deposition of the units with the
1035 doublet feature. Indeed, there is no clear genetic relationship between the Mg/Fe
1036 clay unit and the doublet units, suggesting the latter are inconsistent with in situ
1037 alteration or leaching of the former. One possible interpretation is that the different
1038 proportions of the Al-phyllosilicate component, which correspond to distinct
1039 stratigraphic units, represent changes in sediment flux and or source region,
1040 perhaps where an increase in Al-phyllosilicate corresponds to a more highly
1041 weathered source region. Unlike the other example we presented for this region, the
1042 relative strength of the ~2.27 µm jarosite absorption also appears to conform to
46 1043 stratigraphic boundaries. This may indicate that the jarosite is also detrital, but an
1044 authigenic origin is possible as well since pore fluid flow or chemistry may be
1045 stratigraphically controlled.
1046
1047 If this sequence in Ius Chasma does in fact represent different relative
1048 amounts of authigenic jarosite and detrital Al-phyllosilicate, then we are potentially
1049 observing an upsection transition from precipitate (jarosite) dominated lithologies
1050 to ones dominated by clastic input. Such stratigraphic changes in mineralogy could
1051 be formed in a variety of environments, and without modal abundances for these
1052 minerals it is difficult to narrow the possible interpretations. However, possibilities
1053 include decreasing either the abundance of jarosite cement (which may or may not
1054 correspond to a change in absolute Al-phyllosilicate content), the availability of
1055 water, and/or Fe or S availability. Alternatively, a change in redox conditions, pH, or
1056 an increase in detrital input through time could also influence the relative
1057 proportions of jarosite and Al-phyllosilicate. Regardless, the observation that there
1058 are deposits suggestive of both in situ alteration and a combination of detrital influx
1059 of clay and precipitation of sulfate shows that multiple aqueous processes were
1060 likely responsible for formation of the doublet-bearing materials.
1061
1062 One of the most intriguing features of the doublet-bearing rocks at Rio Tinto
1063 is that the associated mineral assemblage is not preserved through geologic time in
1064 a terrestrial environment. Influx of fresh meteoritic water removes sulfates,
1065 including jarosite, and results in a net increase in ferric oxides, predominantly
47 1066 goethite, through time [Elwood Madden et al., 2004; Golden et al., 2005]. The
1067 transition from young sulfate-rich streambed deposits to older ferric oxide-rich
1068 terraces takes place in less than a million years, which is in agreement with jarosite
1069 dissolution rates reported by Dixon et al. [2015]. That study also noted that jarosite
1070 dissolution rate is more likely to be controlled by mineral-surface reaction rates
1071 rather than hydrologic conditions. Assuming that the Valles Marineris deposits have
1072 been exposed on the surface since at least the Early Amazonian, the apparent
1073 persistence of jarosite indicates that interaction between the doublet materials and
1074 water (e.g., diagenetic overprinting) has been rather limited since the time of
1075 jarosite formation or deposition. Thus, while water is required for the formation of
1076 jarosite and phyllosilicates, and for some occurrences of the doublet materials water
1077 may have acted as a sediment transport agent, it is likely that post-deposition water-
1078 rock interaction has been significantly limited in these regions for hundreds of
1079 millions of years or longer.
1080
1081 In summary, it is important to keep in mind that the presence of a mineral or
1082 mineral assemblage in and of itself is not necessarily indicative of local aqueous
1083 geochemistry for a given outcrop or region. Mineral identifications, particularly
1084 those determined at spatial scales that are typical of orbital spectrometers (10s of
1085 meter per pixel or greater), must be placed in proper context to determine how they
1086 relate to geologic processes and environmental conditions. Specifically, it is critical
1087 to determine which minerals may be detrital versus authigenic (to the extent
1088 possible) and to assess the textural relationships between different mineral phases.
48 1089 Do sulfates and/or phyllosilicates occur as cementing agents, vein fill, or clasts? For
1090 minerals that are indicative of specific chemical conditions, such as jarosite, does the
1091 abundance and/or presence of the phases require those conditions to be persistent
1092 or might they represent short-lived exceptions to very different background
1093 conditions? As it stands, addressing such questions typically requires detailed small-
1094 scale observations that are only achievable by rovers or humans, thus caution
1095 should be exercised when interpreting orbital observations in terms of specific
1096 geologic processes, particularly at small spatial scales.
1097
1098 6. Summary and Conclusions
1099
1100 An enigmatic doublet feature in reflectance spectra has been observed
1101 alongside hydrated mineral deposits in a number of regions on Mars. Previous
1102 studies have proposed that this spectral signature could arise from mixtures of
1103 hydrated minerals with known absorptions between 2.2 and 2.3 µm, including
1104 jarosite, gypsum, silica, Fe/Mg smectite, and Al-phyllosilicates. Possible spectral and
1105 mineralogical analogs from Rio Tinto are analyzed with a combination of laboratory,
1106 field, and aerial reflectance spectra and XRD analysis of the mineralogy and reveal
1107 that the doublet arises in rocks containing both illite or muscovite and jarosite. The
1108 martian doublet materials are also consistent with a mixture of Al-phyllosilicate and
1109 jarosite because (1) the wavelength positions of the martian doublet absorptions
1110 are similar to those observed in Rio Tinto data collected at lab, field, and airborne
1111 spatial scales, (2) the wavelength positions and widths of individual absorptions are
49 1112 consistent with the presence of Al-OH and Fe-OH-bearing minerals, and (3) the
1113 strengths of the two absorptions in the doublet vary independently, implying the
1114 spectra represent a mixture of two or more components. Likely Al-phyllosilicates for
1115 the martian examples include montmorillonite, illite, and kaolinite/halloysite.
1116
1117 Two scenarios are proposed for the creation of these clay-sulfate
1118 assemblages on Mars in the context of a case study of deposits in Ius Chasma. In one
1119 example there are clear mineralogical changes (variations in relative proportions of
1120 jarosite and Al-phyllosilicate) within a single breccia unit. These deposits are
1121 consistent with in situ jarosite formation, possibly linked to diagenetic fluids. In a
1122 second example, observed changes in the relative proportions of the two minerals
1123 appear to follow stratigraphic boundaries, and in all cases the doublet-bearing units
1124 lie unconformably over Mg/Fe clay-bearing strata. A detrital origin for both phases
1125 cannot be ruled out in this case, but it is also plausible that the observed changes in
1126 mineralogy reflect changes in the relative contribution of precipitates (e.g., jarosite)
1127 versus detritus (e.g., Al-phyllosilicates) through time. Both detrital and authigenic
1128 origins are consistent for different doublet-bearing deposits in Ius Chasma, and the
1129 complexity of these deposits demonstrates the need for detailed mapping of each
1130 occurrence to decipher their likely geologic history. The results of such detailed
1131 mapping studies will ultimately provide the most insight into what these deposits
1132 record about aqueous activity, preservation potential, and water-rock-atmosphere
1133 interaction on Mars.
1134
50 1135 Acknowledgements
1136 The authors would like to acknowledge the support of the NASA Astrobiology
1137 Institute (MIT team, award NNA13AA90A) for funding this work. We also thank
1138 John Mustard for providing the HyMap data and the NASA Reflectance Experiment
1139 LABoratory (RELAB) facility for providing some of the spectral data used in this
1140 study. Comments from an anonymous reviewer and Liz Rampe helped to improve
1141 this manuscript and their efforts are graciously acknowledged. All data used in this
1142 paper are available upon request through the corresponding author
1143 ([email protected]).
1144 1145 References 1146 1147 Altheide, T.S., Chevrier, V.F., Noe Dobrea, E., 2010. Mineralogical characterization of acid weathered 1148 phyllosilicates with implications for secondary martian deposits. Geochimica et Cosmochimica 1149 Acta 74, 6232–6248. doi:10.1016/j.gca.2010.08.005 1150 Arvidson, R.E., Ruff, S.W., Morris, R.V., Ming, D.W., Crumpler, L.S., Yen, A.S., Squyres, S.W., Sullivan, 1151 R.J., Bell, J.F., Cabrol, N.A., Clark, B.C., Farrand, W.H., Gellert, R., Greenberger, R., Grant, J.A., 1152 Guinness, E.A., Herkenhoff, K.E., Hurowitz, J.A., Johnson, J.R., Klingelhöfer, G., Lewis, K.W., 1153 Li, R., McCoy, T.J., Moersch, J., McSween, H.Y., Murchie, S.L., Schmidt, M., Schröder, C., 1154 Wang, A., Wiseman, S., Madsen, M.B., Goetz, W., McLennan, S.M., 2008. Spirit Mars Rover 1155 Mission to the Columbia Hills, Gusev Crater: Mission overview and selected results from the 1156 Cumberland Ridge to Home Plate. Journal of Geophysical Research 113. 1157 doi:10.1029/2008JE003183 1158 Bibring, J.-P., 2006. Global Mineralogical and Aqueous Mars History Derived from OMEGA/Mars 1159 Express Data. Science 312, 400–404. doi:10.1126/science.1122659 1160 Bibring, J.-P., Arvidson, R.E., Gendrin, A., Gondet, B., Langevin, Y., Le Mouelic, S., Mangold, N., Morris, 1161 R.V., Mustard, J.F., Poulet, F., Quantin, C., Sotin, C., 2007. Coupled Ferric Oxides and Sulfates 1162 on the Martian Surface. Science 317, 1206–1210. doi:10.1126/science.1144174 1163 Bish, D.L., Howard, S.A., 1988. Quantitative phase analysis using the Rietveld method. Journal of Applied 1164 Crystallography 21, 86–91. doi:10.1107/S0021889887009415 1165 Bishop, J.L., Loizeau, D., McKeown, N.K., Saper, L., Dyar, M.D., Des Marais, D.J., Parente, M., Murchie, 1166 S.L., 2013. What the ancient phyllosilicates at Mawrth Vallis can tell us about possible 1167 habitability on early Mars. Planetary and Space Science 86, 130–149. 1168 doi:10.1016/j.pss.2013.05.006 1169 Bristow, T.F., Bish, D.L., Vaniman, D.T., Morris, R.V., Blake, D.F., Grotzinger, J.P., Rampe, E.B., Crisp, 1170 J.A., Achilles, C.N., Ming, D.W., Ehlmann, B.L., King, P.L., Bridges, J.C., Eigenbrode, J.L., 1171 Sumner, D.Y., Chipera, S.J., Moorokian, J.M., Treiman, A.H., Morrison, S.M., Downs, R.T., 1172 Farmer, J.D., Marais, D.D., Sarrazin, P., Floyd, M.M., Mischna, M.A., McAdam, A.C., 2015. The 1173 origin and implications of clay minerals from Yellowknife Bay, Gale crater, Mars. American 1174 Mineralogist 100, 824–836. doi:10.2138/am-2015-5077CCBYNCND 1175 Burns, R.G., 1993. Rates and mechanisms of chemical weathering of ferromagnesian silicate minerals on 1176 Mars. Geochimica et Cosmochimica Acta 57, 4555–4574. doi:10.1016/0016-7037(93)90182-V
51 1177 Burns, R.G., 1987. Gossans on Mars. 18th Lunar and Planetary Science Conference 713–721. 1178 Carter, J., Poulet, F., Bibring, J.-P., Mangold, N., Murchie, S., 2013. Hydrous minerals on Mars as seen by 1179 the CRISM and OMEGA imaging spectrometers: Updated global view: HYDROUS MINERALS 1180 ON MARS: GLOBAL VIEW. Journal of Geophysical Research: Planets 118, 831–858. 1181 doi:10.1029/2012JE004145 1182 Cavanagh, P.D., Bish, D.L., Blake, D.F., Vaniman, D.T., Morris, R.V., Ming, D.W., Rampe, E.B., Achilles, 1183 C.N., Chipera, S.J., Treiman, A.H., Downs, R.T., Morrison, S.M., Fendrich, K.V., Yen, A.S., 1184 Grotzinger, J., Crisp, J.A., Bristow, T.F., Sarrazin, P., Farmer, J.D., Des Marais, Stolper, E., 1185 Morookian, J.M., Wilson, M.A., Spanovich, N., Anderson, R., 2015. Confidence Hills Mineralogy 1186 and CheMin Results from Base of Mt. Sharp, Pahrump Hills, Gale Crater, Mars. 46th Lunar and 1187 Planetary Science Conference Abstract #2735. 1188 Chemtob, S.M., Arvidson, R.E., FernáNdez-Remolar, D., Amils, R., 2006. Identification of Hydrated 1189 Sulfates Collected in the Northern Rio Tinto Valley by Reflectance and Raman Spectroscopy. 1190 Lunar and Planetary Science XXXVII Abstract 1941. 1191 Clark, R.N., Roush, T.L., 1984. Reflectance spectroscopy: Quantitative analysis techniques for remote 1192 sensing applications. Journal of Geophysical Research 89, 6329. doi:10.1029/JB089iB07p06329 1193 Dixon, E.M., Elwood Madden, A.S., Hausrath, E.M., Elwood Madden, M.E., 2015. Assessing 1194 hydrodynamic effects on jarosite dissolution rates, reaction products, and preservation on Mars: 1195 Hydrodynamics effects Mars Minerals. Journal of Geophysical Research: Planets 120, 625–642. 1196 doi:10.1002/2014JE004779 1197 Ehlmann, B.L., Mustard, J.F., Murchie, S.L., Bibring, J.-P., Meunier, A., Fraeman, A.A., Langevin, Y., 1198 2011. Subsurface water and clay mineral formation during the early history of Mars. Nature 479, 1199 53–60. doi:10.1038/nature10582 1200 Ehlmann, B.L., Mustard, J.F., Swayze, G.A., Clark, R.N., Bishop, J.L., Poulet, F., Des Marais, D.J., Roach, 1201 L.H., Milliken, R.E., Wray, J.J., Barnouin-Jha, O., Murchie, S.L., 2009. Identification of hydrated 1202 silicate minerals on Mars using MRO-CRISM: Geologic context near Nili Fossae and implications 1203 for aqueous alteration. Journal of Geophysical Research 114. doi:10.1029/2009JE003339 1204 Elwood Madden, M.E., Bodnar, R.J., Rimstidt, J.D., 2004. Jarosite as an indicator of water-limited 1205 chemical weathering on Mars. Nature 431, 821–823. doi:10.1038/nature02971 1206 Farrand, W.H., Glotch, T.D., Rice, J.W., Hurowitz, J.A., Swayze, G.A., 2009. Discovery of jarosite within 1207 the Mawrth Vallis region of Mars: Implications for the geologic history of the region. Icarus 204, 1208 478–488. doi:10.1016/j.icarus.2009.07.014 1209 Fernández-Remolar, D.C., 2003. Geological record of an acidic environment driven by iron 1210 hydrochemistry: The Tinto River system. Journal of Geophysical Research 108. 1211 doi:10.1029/2002JE001918 1212 Fernández-Remolar, D.C., Knoll, A.H., 2008. Fossilization potential of iron-bearing minerals in acidic 1213 environments of Rio Tinto, Spain: Implications for Mars exploration. Icarus 194, 72–85. 1214 doi:10.1016/j.icarus.2007.10.009 1215 Fernández-Remolar, D.C., Morris, R.V., Gruener, J.E., Amils, R., Knoll, A.H., 2005. The Río Tinto Basin, 1216 Spain: Mineralogy, sedimentary geobiology, and implications for interpretation of outcrop rocks at 1217 Meridiani Planum, Mars. Earth and Planetary Science Letters 240, 149–167. 1218 doi:10.1016/j.epsl.2005.09.043 1219 Fernández-Remolar, D.C., Prieto-Ballesteros, O., Gómez-Ortíz, D., Fernández-Sampedro, M., Sarrazin, P., 1220 Gailhanou, M., Amils, R., 2011. Río Tinto sedimentary mineral assemblages: A terrestrial 1221 perspective that suggests some formation pathways of phyllosilicates on Mars. Icarus 211, 114– 1222 138. doi:10.1016/j.icarus.2010.09.008 1223 Fialips, C.-I., Petit, S., Decarreau, A., Beaufort, D., 2000. Influence of synthesis pH on kaolinite 1224 “crystallinity” and surface properties. Clays and Clay Minerals 48, 173–184. 1225 Gendrin, A., 2005. Sulfates in Martian Layered Terrains: The OMEGA/Mars Express View. Science 307, 1226 1587–1591. doi:10.1126/science.1109087 1227 Golden, D.C., Ming, D.W., Morris, R.V., Mertzman, S.A., 2005. Laboratory-simulated acid-sulfate 1228 weathering of basaltic materials: Implications for formation of sulfates at Meridiani Planum and 1229 Gusev crater, Mars. Journal of Geophysical Research 110. doi:10.1029/2005JE002451 1230 Graff, T.G., Morris, R.V., Achilles, C.N., Agresti, A.G., Ming, D.W., Hamilton, J.C., Mertzman, S.A., 1231 Smith, J., 2012. Chemical and Mineralogical Characterization of Acid-Sulfate Alteration of
52 1232 Basaltic Material on Mauna Kea Volcano, Hawaii: Jarosite and Hydrated Halloysite. Lunar and 1233 Planetary Science XLIII Abstract, 2639. 1234 Grotzinger, J.P., Arvidson, R.E., Bell, J.F., Calvin, W., Clark, B.C., Fike, D.A., Golombek, M., Greeley, 1235 R., Haldemann, A., Herkenhoff, K.E., Jolliff, B.L., Knoll, A.H., Malin, M., McLennan, S.M., 1236 Parker, T., Soderblom, L., Sohl-Dickstein, J.N., Squyres, S.W., Tosca, N.J., Watters, W.A., 2005. 1237 Stratigraphy and sedimentology of a dry to wet eolian depositional system, Burns formation, 1238 Meridiani Planum, Mars. Earth and Planetary Science Letters 240, 11–72. 1239 doi:10.1016/j.epsl.2005.09.039 1240 Grotzinger, J.P., Gupta, S., Malin, M.C., Rubin, D.M., Schieber, J., Siebach, K., Sumner, D.Y., Stack, 1241 K.M., Vasavada, A.R., Arvidson, R.E., Calef, F., Edgar, L., Fischer, W.F., Grant, J.A., Griffes, J., 1242 Kah, L.C., Lamb, M.P., Lewis, K.W., Mangold, N., Minitti, M.E., Palucis, M., Rice, M., 1243 Williams, R.M.E., Yingst, R.A., Blake, D., Blaney, D., Conrad, P., Crisp, J., Dietrich, W.E., 1244 Dromart, G., Edgett, K.S., Ewing, R.C., Gellert, R., Hurowitz, J.A., Kocurek, G., Mahaffy, P., 1245 McBride, M.J., McLennan, S.M., Mischna, M., Ming, D., Milliken, R., Newsom, H., Oehler, D., 1246 Parker, T.J., Vaniman, D., Wiens, R.C., Wilson, S.A., 2015. Deposition, exhumation, and 1247 paleoclimate of an ancient lake deposit, Gale crater, Mars. Science 350, aac7575–aac7575. 1248 doi:10.1126/science.aac7575 1249 Grotzinger, J.P. et al., 2014. A Habitable Fluvio-Lacustrine Environment at Yellowknife Bay, Gale Crater, 1250 Mars. Science 343, 1242777–1242777. doi:10.1126/science.1242777 1251 Hurowitz, J.A., Fischer, W.W., Tosca, N.J., Milliken, R.E., 2010. Origin of acidic surface waters and the 1252 evolution of atmospheric chemistry on early Mars. Nature Geoscience 3, 323–326. 1253 doi:10.1038/ngeo831 1254 Hurowitz, J.A., Tosca, N.J., Dyar, M.D., 2009. Acid production by FeSO4{middle dot}nH2O dissolution 1255 and implications for terrestrial and martian aquatic systems. American Mineralogist 94, 409–414. 1256 doi:10.2138/am.2009.3085 1257 Klingelhofer, G., 2004. Jarosite and Hematite at Meridiani Planum from Opportunity’s Mossbauer 1258 Spectrometer. Science 306, 1740–1745. doi:10.1126/science.1104653 1259 Langevin, Y., 2005. Sulfates in the North Polar Region of Mars Detected by OMEGA/Mars Express. 1260 Science 307, 1584–1586. doi:10.1126/science.1109091 1261 Lichtenberg, K.A., Arvidson, R.E., Morris, R.V., Murchie, S.L., Bishop, J.L., Fernandez Remolar, D., 1262 Glotch, T.D., Noe Dobrea, E., Mustard, J.F., Andrews-Hanna, J., Roach, L.H., 2010. Stratigraphy 1263 of hydrated sulfates in the sedimentary deposits of Aram Chaos, Mars. Journal of Geophysical 1264 Research 115. doi:10.1029/2009JE003353 1265 Madejová, J., Pentrák, M., Pálková, H., Komadel, P., 2009. Near-infrared spectroscopy: A powerful tool in 1266 studies of acid-treated clay minerals. Vibrational Spectroscopy 49, 211–218. 1267 doi:10.1016/j.vibspec.2008.08.001 1268 Metz, J., Grotzinger, J., Okubo, C., Milliken, R., 2010. Thin-skinned deformation of sedimentary rocks in 1269 Valles Marineris, Mars. Journal of Geophysical Research 115. doi:10.1029/2010JE003593 1270 Metz, J.M., Grotzinger, J.P., Mohrig, D., Milliken, R., Prather, B., Pirmez, C., McEwen, A.S., Weitz, C.M., 1271 2009. Sublacustrine depositional fans in southwest Melas Chasma. Journal of Geophysical 1272 Research 114. doi:10.1029/2009JE003365 1273 Michalski, J.R., Niles, P.B., Cuadros, J., Baldridge, A.M., 2013. Multiple working hypotheses for the 1274 formation of compositional stratigraphy on Mars: Insights from the Mawrth Vallis region. Icarus 226, 1275 816–840. doi:10.1016/j.icarus.2013.05.024 1276 Milliken, R.E., Swayze, G.A., Arvidson, R.E., Bishop, J.L., Clark, R.N., Ehlmann, B.L., Green, R.O., 1277 Grotzinger, J.P., Morris, R.V., Murchie, S.L., Mustard, J.F., Weitz, C., 2008. Opaline silica in 1278 young deposits on Mars. Geology 36, 847. doi:10.1130/G24967A.1 1279 Murchie, S., Arvidson, R., Bedini, P., Beisser, K., Bibring, J.-P., Bishop, J., Boldt, J., Cavender, P., Choo, 1280 T., Clancy, R.T., Darlington, E.H., Des Marais, D., Espiritu, R., Fort, D., Green, R., Guinness, E., 1281 Hayes, J., Hash, C., Heffernan, K., Hemmler, J., Heyler, G., Humm, D., Hutcheson, J., Izenberg, 1282 N., Lee, R., Lees, J., Lohr, D., Malaret, E., Martin, T., McGovern, J.A., McGuire, P., Morris, R., 1283 Mustard, J., Pelkey, S., Rhodes, E., Robinson, M., Roush, T., Schaefer, E., Seagrave, G., Seelos, 1284 F., Silverglate, P., Slavney, S., Smith, M., Shyong, W.-J., Strohbehn, K., Taylor, H., Thompson, 1285 P., Tossman, B., Wirzburger, M., Wolff, M., 2007. Compact Reconnaissance Imaging 1286 Spectrometer for Mars (CRISM) on Mars Reconnaissance Orbiter (MRO). Journal of Geophysical 1287 Research 112. doi:10.1029/2006JE002682
53 1288 Mustard, J.F., Murchie, S.L., Pelkey, S.M., Ehlmann, B.L., Milliken, R.E., Grant, J.A., Bibring, J.-P., 1289 Poulet, F., Bishop, J., Dobrea, E.N., Roach, L., Seelos, F., Arvidson, R.E., Wiseman, S., Green, 1290 R., Hash, C., Humm, D., Malaret, E., McGovern, J.A., Seelos, K., Clancy, T., Clark, R., Marais, 1291 D.D., Izenberg, N., Knudson, A., Langevin, Y., Martin, T., McGuire, P., Morris, R., Robinson, 1292 M., Roush, T., Smith, M., Swayze, G., Taylor, H., Titus, T., Wolff, M., 2008. Hydrated silicate 1293 minerals on Mars observed by the Mars Reconnaissance Orbiter CRISM instrument. Nature 454, 1294 305–309. doi:10.1038/nature07097 1295 Noe Dobrea, E., Michalski, J.R., Swayze, G.A., 2011. Aqueous Mineralogy and Stratigraphy at and Around 1296 the Proposed Mawrth Vallis MSL Landing Site: New Insights into the Aqueous History of the 1297 Region. The International Journal of Mars Science and Exploration 6, 32–46. 1298 Pelkey, S.M., Mustard, J.F., Murchie, S., Clancy, R.T., Wolff, M., Smith, M., Milliken, R., Bibring, J.-P., 1299 Gendrin, A., Poulet, F., Langevin, Y., Gondet, B., 2007. CRISM multispectral summary products: 1300 Parameterizing mineral diversity on Mars from reflectance. Journal of Geophysical Research 112. 1301 doi:10.1029/2006JE002831 1302 Poulet, F., Bibring, J.-P., Mustard, J.F., Gendrin, A., Mangold, N., Langevin, Y., Arvidson, R.E., Gondet, 1303 B., Gomez, C., Berthé, M., Bibring, J.-P., Langevin, Y., Erard, S., Forni, O., Gendrin, A., Gondet, 1304 B., Manaud, N., Poulet, F., Poulleau, G., Soufflot, A., Combes, M., Drossart, P., Encrenaz, T., 1305 Fouchet, T., Melchiorri, R., Bellucci, G., Altieri, F., Formisano, V., Fonti, S., Capaccioni, F., 1306 Cerroni, P., Coradini, A., Korablev, O., Kottsov, V., Ignatiev, N., Titov, D., Zasova, L., Mangold, 1307 N., Pinet, P., Schmitt, B., Sotin, C., Hauber, E., Hoffmann, H., Jaumann, R., Keller, U., Arvidson, 1308 R., Mustard, J., Forget, F., 2005. Phyllosilicates on Mars and implications for early martian 1309 climate. Nature 438, 623–627. doi:10.1038/nature04274 1310 Roach, L.H., Mustard, J., Gendrin, A., Fernández-Remolar, D., Amils, R., Amaral-Zettler, L., 2006. 1311 Finding mineralogically interesting targets for exploration from spatially coarse visible and near 1312 IR spectra. Earth and Planetary Science Letters 252, 201–214. doi:10.1016/j.epsl.2006.09.044 1313 Roach, L.H., Mustard, J.F., Swayze, G., Milliken, R.E., Bishop, J.L., Murchie, S.L., Lichtenberg, K., 2010. 1314 Hydrated mineral stratigraphy of Ius Chasma, Valles Marineris. Icarus 206, 253–268. 1315 doi:10.1016/j.icarus.2009.09.003 1316 Sobron, P., Bishop, J.L., Blake, D.F., Chen, B., Rull, F., 2014. Natural Fe-bearing oxides and sulfates from 1317 the Rio Tinto Mars analog site: Critical assessment of VNIR reflectance spectroscopy, laser 1318 Raman spectroscopy, and XRD as mineral identification tools. American Mineralogist 99, 1199– 1319 1205. doi:10.2138/am.2014.4595 1320 Sobron, P., Sanz, A., Acosta, T., Rull, F., 2009. A Raman spectral study of stream waters and efflorescent 1321 salts in Rio Tinto, Spain. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 1322 71, 1678–1682. doi:10.1016/j.saa.2008.06.035 1323 Squyres, S.W., 2004. In Situ Evidence for an Ancient Aqueous Environment at Meridiani Planum, Mars. 1324 Science 306, 1709–1714. doi:10.1126/science.1104559 1325 Sunshine, J.M., Pieters, C.M., 1993. Estimating Modal Abundances From the Spectra of Natural and 1326 Laboratory Pyroxene Mixtures Using the Modified Gaussian Model. Journal of Geophysical 1327 Research 98, 9075–9087. 1328 Thollot, P., Mangold, N., Ansan, V., Le Mouélic, S., Milliken, R.E., Bishop, J.L., Weitz, C.M., Roach, 1329 L.H., Mustard, J.F., Murchie, S.L., 2012. Most Mars minerals in a nutshell: Various alteration 1330 phases formed in a single environment in Noctis Labyrinthus: ALTERATION PHASES IN 1331 NOCTIS LABYRINTHUS. Journal of Geophysical Research: Planets 117, n/a–n/a. 1332 doi:10.1029/2011JE004028 1333 Tosca, N.J., 2004. Acid-sulfate weathering of synthetic Martian basalt: The acid fog model revisited. 1334 Journal of Geophysical Research 109. doi:10.1029/2003JE002218 1335 Tosca, N.J., Knoll, A.H., McLennan, S.M., 2008a. Water Activity and the Challenge for Life on Early 1336 Mars. Science 320, 1204–1207. doi:10.1126/science.1155432 1337 Tosca, N.J., McLennan, S.M., 2006. Chemical divides and evaporite assemblages on Mars. Earth and 1338 Planetary Science Letters 241, 21–31. doi:10.1016/j.epsl.2005.10.021 1339 Tosca, N.J., Milliken, R.E., Michel, F.M., 2008b. Smectite Formation on Early Mars: Experimental 1340 Constraints. Martian Phyllosilicates: Recorders of Aqueous Processes Abstract, 7030. 1341 Vaniman, D.T. et al., 2014. Mineralogy of a Mudstone at Yellowknife Bay, Gale Crater, Mars. Science 1342 343, 1243480–1243480. doi:10.1126/science.1243480
54 1343 Wang, A., Korotev, R.L., Jolliff, B.L., Haskin, L.A., Crumpler, L., Farrand, W.H., Herkenhoff, K.E., de 1344 Souza, P., Kusack, A.G., Hurowitz, J.A., Tosca, N.J., 2006. Evidence of phyllosilicates in Wooly Figure1345 1 Patch, an altered rock encountered at West Spur, Columbia Hills, by the Spirit rover in Gusev 1346 crater, Mars. Journal of Geophysical Research 111. doi:10.1029/2005JE002516 1347 Weitz, C.M., Bishop, J.L., Thollot, P., Mangold, N., Roach, L.H., 2011. Diverse mineralogies in two 1348 troughs of Noctis Labyrinthus, Mars. Geology 39, 899–902. doi:10.1130/G32045.1 1349 Weitz, C.M., Noe Dobrea, E., Wray, J.J., 2014. Mixtures of clays and sulfates within deposits in western 1350 Melas Chasma, Mars. Icarus. doi:10.1016/j.icarus.2014.04.009 1351 1352 1353 Figures
RELAB spectrum: Muscovite (C1XT09), Montmorillonite (CJS125), Gypsum (C1CC35), Nontronite (C1CY20), Opal (C1RM34). USGS spectrum: Jarosite (GDS98K) 1354 1355 Figure 1: Reflectance spectra of minerals with absorptions between 2.2 and 2.3 μm 1356 (source: RELAB database) are compared to CRISM spectra showing the ‘doublet’ 1357 absorption. Continuum-removed spectra from 2.1 to 2.35 μm (right) are shown to 1358 highlight absorption features in the wavelength region where the doublet occurs. 1359 None of the pure mineral spectra have absorption positions and depths that are a 1360 direct match to the doublet absorptions, thus a mineral mixture is implied. 1361
55 Spectral' Reitveld' Sample'ID Location Terrace Doublet Refinement RT1 Berrocal Modern RT2 Berrocal Modern RT3 Berrocal Modern RT41(mud,1loose) Berrocal Modern RT5 Anabel's1Garden,1Upper1Road N/A RT6 Anabel's1Garden,1Upper1Road N/A RT7 Anabel's1Garden,1Upper1Road N/A Yes RT8 Anabel's1Garden,1Upper1Road N/A RT9 Anabel's1Garden,1Upper1Road N/A Yes Yes RT10 Anabel's1Garden,1Upper1Road N/A Yes Yes RT11 Anabel's1Garden,1Upper1Road N/A Yes Yes RT12 Barranco1de1los1Locos Lower1Terrace Yes Yes RT13 Barranco1de1los1Locos Lower1Terrace RT14 Barranco1de1los1Locos Lower1Terrace RT15 Barranco1de1los1Locos Lower1Terrace Yes Yes RT16 Anabel's1Garden,1Stream11 Modern Yes Yes RT171(precipitate) Anabel's1Garden,1Stream12 Modern RT181(precipitate) Anabel's1Garden,1Stream12 Modern RT19 Anabel's1Garden,1Stream12 Modern Yes Yes RT201(precipitate) Anabel's1Garden,1Stream12 Modern RT21 Anabel's1Garden,1Stream12 Modern RT22 Anabel's1Garden,1Stream12 Modern RT23 Anabel's1Garden,1Stream12 Modern Yes RT24 Source1Spring Modern RT25 Source1Spring Modern RT26 Source1Spring Modern Yes Yes RT271(precipitate) Source1Spring Modern RT28 Alta1de1la1Mesa Upper1Terrace RT29 Alta1de1la1Mesa Upper1Terrace RT30 Alta1de1la1Mesa Upper1Terrace Yes RT31 Alta1de1la1Mesa Upper1Terrace RT32 Gossan1outcrop N/A RT33 Gossan1outcrop N/A 1362 RT34 Gossan1outcrop N/A 1363 1364 Table 1: A full list of samples collected at the modern streambed, lower and upper 1365 terraces at Rio Tinto. All samples are rocks unless otherwise noted. Each sample is 1366 measured with the ASD, FTIR, and XRD in the laboratory. Samples with a spectral 1367 doublet and those that have been analyzed with a Rietveld refinement are noted. 1368
56 1369 1370 Figure 2: Terraces of the Rio Tinto in the schematic diagram from Fernandez- 1371 Remolar et al. (2005) are matched with locations in the HyMap aerial hyperspectral 1372 image of the region that were sampled in this study. 1373
57 1374
1375 1376 Figure 3: A principal components analysis has been carried out on the same HyMap 1377 image as seen in Figure 2, and like colors in the image correspond to regions of 1378 spectral similarity (R = PCA band 7, G = band 3, B = band 5). The spectral doublet is 1379 associated with units that appear in a dark red color in this false-color image, which 1380 is particularly evident at Anabel’s Garden (inset) where a number of the doublet- 1381 bearing samples were collected. Variation in the spectral doublet from 2.2 – 2.3 μm 1382 is highlighted with spectra collected from a 500 m HyMap transect (black arrow). 1383 1384 Sample Rietveld4Results XRD4Peak4Areas Jarosite4 Muscovite4 Illite4 Peak4area:4 Peak4area:4 Sample Location Doublet? R (wt%) (wt%) (wt%) wp Musc/Ill Jarosite RT9$ AG$Mining$Terrace Yes 3 16 14 5.61 1717.8 639 RT10 AG$Mining$Terrace Yes 7 46 0 5.01 2438.3 673.4 RT11 AG$Mining$Terrace Yes 4 24 0 5.52 1239.4 573.1 RT12$ Barranco$de$los$Locos Yes 11 13 0 1.12 2115.3 1956.1 RT15$ Barranco$de$los$Locos Yes 37 12 0 1.21 2089.9 2050.9 RT16 Anabel's$Garden$(Stream) Yes 3 5 0 10.08 555.4 353.4 RT19 Anabel's$Garden$(stream) Yes 3 0 19 3.16 536.3 464.9 RT26 Source$Spring Yes 8 50 0 2.52 4655.9 2959.1 RT30 Alta$de$la$Mesa No 0 0 0 3.67 0 0 RT23 Anabel's$Garden$(Stream) No 32 0 0 1.66 0 1003.8 1385 RT7 Anabel's$Garden No 0 0 4 1.03 432.1 0
58 1386 Table 2: Sample IDs and origin of samples that were analyzed using XRD and 1387 reflectance spectroscopy. The jarosite and Al-clay (muscovite/illite) content of each 1388 sample was estimated using Rietveld refinemenFigure 4 t on the XRD patterns. 1389 1390
Band Centers 2000
1500
1000
Noise Counts 0.015 500 0 2.1 2.2 2.3 0.01 Wavelength of Band Center
Noise 0.005 2500 2000 0 1500 Counts 1000
500
1 km 0 0 0.005 0.01 0.015 Standard Deviation (from 1.7 - 1.85 μm)
1391 1392 Figure 4: Two Gaussian fit examples are shown for FRT00009B27 from Ius Chasma, 1393 Mars. Due to noise in the spectrum, the Gaussian mean may be a better 1394 representative of band depth minimum than the true reflectance value. Noise and 1395 spectral smile are further analyzed for a central subset of the image. Noise is 1396 calculated as the standard deviation of a flat, absorption free portion of the 1397 spectrum from 1.7 – 1.85 µm. An image and histogram of this standard deviation 1398 show that noise is typically < 0.005 in reflectance. Band center histograms are also 1399 shown for the central image subset, and the reduced spread in band centers 1400 compared to the full image suggests spectral smile plays a role in the band center 1401 position. 1402
59 Figure 5 1403
A B
C
1404 1405 Figure 5: Example spectra from the field and laboratory for the A) modernly 1406 precipitating streambed at Rio Tinto (source region, which includes Source Spring, 1407 Richi Spring, and Anabel’s Garden) and B) the ~2myr old terrace at Alto de la Mesa. 1408 These spectra show the transition from very hydrated, sulfate minerals to ferric 1409 oxides through time (with some clays possibly associated with mining activity). All 1410 of the mineralogies identified in hand samples are also found in the field spectra, but 1411 only a limited number of each is shown for the sake of clarity. In C), absorptions in 1412 powdered samples that show a doublet are specifically compared to jarosite 1413 spectral endmembers. Again, a very similar doublet signature is seen in both field 1414 and hand samples, but is not shown for clarity. Endmember mineral spectra from 1415 RELAB and USGS are shown for reference (RELAB: RM-JFM-036, RM-JFM-035, CC- 1416 JFM-013A, JB-JLB-752, JB-JLB-753, JB-JLB-757, JB-CMP-047, JB-CMP-054, CC-JFM-
60 1417 017-B, JB-JLB_256, CC-JFM-035, CL-TXH-014, JB-CMP-25-B; USGS: BZ93-1, GDS2 Figure 5 1418 Zeolite W1R1Bb) 1419 Jarosite (%)/Al-clay (%) = 0.17 Sample Fit Error
Jarosite (%)/Al-clay (%) = 0.74
Jarosite (%)/Al-clay (%) = 3.15
Goethite Muscovite/ Jarosite Quartz 1420 Illite 1421 Figure 6: XRD patterns with corresponding reflectance spectra from three 1422 powdered samples with different proportions of jarosite and Al-clay (RT 11, RT15, 1423 RT16). Major phases are labeled for the XRD patterns and absorptions 1424 corresponding to the doublet feature are marked by red arrows in the reflectance 1425 spectra plots. 1426
61 RT19 RT10
Figure 8 1427 1428 Figure 7: Example reflectance spectra acquired in the field and for hand sample, 1429 chip, and powder versions of two samples from different locations within the study 1430 region. Chips are <9 mm in diameter and powders were sieved to <45 µm. Spectra of 1431 these samples highlight the variation in absolute reflectance and band depth of the 1432 two absorptions in the doublet feature when measured at different spatial scales. 1433 Hand sample RT10 has a varnished surface for which the reflectance spectra exhibit 1434 a clear doublet feature, whereas spectra from the interior portions exhibit a much 1435 weaker long-wavelength (jarosite) absorption. 1436
Relative Band Depth vs. Al-Clay vs. Band Depth 1 1 XRD Peak Area 100
RT10 2 2 0.8 R = 0.75236 80 R = 0.48937 RT11 0.6 60
RT9 0.4 40 Al-clay (wt%) 0.2 20
0 0 0 0.2 0.4 0.6 0.8 1 0 0.02 0.04 0.06 0.08 RT26 Al−clay / (Al−clay+Jarosite) Band Depth at 2.205 microns Relative Band Depth vs. Jarosite vs. Band Depth 2 1 XRD Reitveld Refinement RT16 100 R2 = 0.75446 0.8 2 RT19 80 R = 0.85938
RT12 0.6 60 RT15 0.4 40
0.2 Jarosite (wt%) 20
0 0 0 0.2 0.4 0.6 0.8 1 0 0.02 0.04 0.06 0.08 Al−clay / (Al−clay+Jarosite) Band Depth at 2.269 microns Band Depth 1/ (Band Depth 1 + 2) 1437
62 1438 Figure 8: Comparison of XRD results with reflectance spectra (band depths) to 1439 show relationship between mineral abundances and the absorption feature 1440 strength. A direct relationship between relative amounts of jarosite and Al-clay and 1441 the strengths of the absorption features indicates the shorter wavelength 1442 absorption (band 1; ~2.21 µm) arises from Al-clay and the longer wavelength 1443 absorption (band 2; ~2.265) arises from the presence of jarosite. This suggests that 1444 relative band strengths are directly related to the relative proportions of the 1445 phyllosilicate and sulfate components. 1446 Image Band)1)(µm) 2σ)(µm) Band)2)(µm) 2σ)(µm) Mars FRT00009B27)full)image 2.2109 0.0264 2.2709 0.019 FRT00009B27)center 2.2006 0.0134 2.265 0.0112 FRT0000A396)full)image 2.2048 0.0154 2.2696 0.0132 FRT0000A396)center 2.2055 0.0134 2.2705 0.0136 FRT0000AD3D)full)image 2.2055 0.0188 2.2662 0.0156 FRT0000AD3D)center 2.2085 0.0142 2.2671 0.0108 Rio)Tinto 1447 HyMap 2.2025 0.013 2.2602 0.011 1448 Table 3: The results of Gaussian fits for every pixel in three CRISM images (Mars) 1449 and the HyMap image (Rio Tinto) that includes the mean band centers and the 2σ 1450 deviations on these band centers. A central subset of the CRISM images is also 1451 analyzed and reported results show a shift in band centers and a decrease in 1452 standard deviation.
63 1453 Figure 9
A 50 km FRT00009B27 FRT0000A396 FRT0000AD3D
5 km 5 km 5 km
B
4 4 x10 Band Centers x10 Band Centers 4 Band Centers C 6 8 9x10 5 7 8 6 7 4 5 6 3 4 5
Counts 4 Counts
Counts 3 2 3 2 1 2 1 1 0 0 0 2.15 2.2 2.25 2. 3 2.15 2.2 2.25 2. 3 2.15 2.2 2.25 2.3 Wavelength of Band Center Wavelength of Band Center Wavelength of Band Center
Relative Amplitude Relative Amplitude Relative Amplitude 4 4 x104 of Absorptions x10 of Absorptions x10 of Absorptions D 2 2.5 2.5 Jarosite Al-Clay 2 Rich Rich 1 Counts
1 Counts 1 Counts 0 0 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 Band 1 Amp./ Band 2 Amp. Band 1 Amp./ Band 2 Amp. Band 1 Amp./ Band 2 Amp. 1454 1455 Figure 9: Examples of doublet-bearing materials on Mars as observed in MRO 1456 CRISM images. (A) MOLA hillshade image showing the location of three study 1457 regions in Ius and Melas Chasma within the Valles Marineris. Zoom-in regions below 1458 show locations of CRISM images FRT00009B27, FRT0000A396, and FRT0000AD3D 1459 in context with nearby CRISM images (overlain on HiRISE and CTX images). 1460 Doublet-bearing units in CRISM images 9B27 and AD3D are light toned and tend to 1461 drape pre-existing topography; CRISM image A396 shows that the doublet-bearing
64 1462 material corresponds to a blocky, brecciated unit. (B) Examples of spectral variation 1463 within the doublet-bearing units, ranging from a stronger 2.205 μm absorption to a 1464 stronger 2.269 μm absorption. (C) Histograms showing positions of the 2.205 μm 1465 (blue) and 2.269 μm (red) band positions for every doublet-bearing pixel in each of 1466 the three images and (D) histograms showing the relative strength of the two 1467 absorption features (Band 1 = ~2.205 µm; Band 2 = ~2.269 µm). Values in (C) and 1468 (D) are based on Gaussian fits as described in the text. Data for each column in (B), 1469 (C), and (D) correspond to the CRISM FRT images labeled and highlighted in red in 1470 (A). 1471
1472 1473 Figure 10: Examples of doublet-bearing units as observed in MRO HiRISE images. 1474 These units are typically light-toned, contorted and/or blocky in nature. (A) 1475 Example of a doublet-bearing unit in Melas Chasma that drapes pre-existing 1476 topography and spans a wide range in elevation. (B) An example of doublet-bearing, 1477 light-toned units that may have slumped off the canyon walls and are now part of a 1478 brecciated unit on the floor. The morphology and spectral characteristics of these 1479 and similar deposits are described in detail in Roach et al. (2010), Metz et al. (2010), 1480 and Weitz et al. (2014). (HiRISE images: (A) PSP_007074_1725, (B) 1481 PSP_007509_1720).
65 1482
1483 1484 Figure 11: Relative amplitude of absorptions (band depth at 2.207/ band depth at 1485 2.269) in Ius Chasma and associated morphology. (A) False-color CRISM images (R – 1486 band 253, B – band 53, G = band 17, top) overlain on a CTX mosaic. Fe/Mg-clay 1487 bearing regions appear greenish in color and the doublet-bearing regions appear 1488 light-toned or white. (B) CRISM pixels with the doublet feature are highlighted by 1489 different colors that map the relative band strength of the two absorptions, with a 1490 gradient from more jarosite rich (red), to roughly equal jarosite/Al-phyllosilicate 1491 strength, to slightly more Al-phyllosilicate rich (blue), to very Al-phyllosilicate rich 1492 (yellow). (C) Close-up view of HiRISE image ESP_041121_1725 showing that there is 1493 no clear stratigraphic boundary associated with the transition from a more jarosite-
66 1494 rich “red” region to a more Al-phyllosilicate rich “blue” area in this location. (D) 1495 Close-up of HiRISE image PSP_007074_1725 showing an example where 1496 morphologically and stratigraphically distinct units correspond to the different 1497 relative band strengths. Here, white arrows indicate a clear stratigraphic boundary 1498 that also corresponds to a clear boundary between green and blue pixels seen in (B) 1499 (HiRISE images: ESP_041121_1725, PSP_007074_1725, CRISM images listed above). 1500
1501 1502 Figure 12: Evidence of the stratigraphic contacts between different doublet-bearing 1503 units that are identified as more jarosite- or Al-phyllosilicate-rich in the color map in 1504 Figure 11. (A) Context map from HiRISE image ESP_041121_1725 showing locations 1505 where three contacts are visible. (B) Shows the clay bearing unit (dark-toned, 1506 stratified with tilted bedding) unconformably overlain by the light-toned doublet 1507 unit, which has relative band strengths between 0.9 and 1.2 (i.e. the strength of the 1508 two absorptions in the doublet are similar). Next, in (C) there is a contact between 1509 this same light-toned unit and a slightly darker, more Al-phyllosilicate-rich unit, 1510 which appears to lie stratigraphically above it. Finally, (D) shows the contact 1511 between the brightest unit, which outcrops in smaller areas throughout Ius, and the
67 1512 light-toned doublet-bearing unit shown in (B) and (C). This bright unit is mapped as 1513 being more jarosite rich in Figure 8 and appears to be stratigraphically lower than 1514 the light-toned unit. It is not possible to determine the exact relationship between 1515 this bright material and the dark, stratified clay deposit in this location, but in other 1516 regions it appears as if the brightest (jarosite-rich) material is stratigraphically 1517 above the clay-bearing strata. Therefore, the increasing stratigraphic order appears 1518 to be: dark (clay-bearing) strata > brightest unit > light-toned unit > moderate- 1519 toned unit. The corresponding mineralogy is consistent with: clay-rich > more 1520 jarosite-rich > equal Al-phyllosilicate and jarosite > more Al-phyllosilicate rich (that 1521 is, within the doublet-bearing units the proportion of jarosite decreases upsection). 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 Supplementary Material 1535 1536 1537 1538 1. Methods 1539 1540 Additional information from the methods section can be found here. 1541 1542 1543 1.1. CRISM image processing and Gaussian fitting 1544 1545 CRISM image cubes were corrected for atmospheric gases using the ‘volcano
1546 scan’ method described by Langevin et al. [2005] and Mustard et al. [2008]. All
1547 CRISM processing steps, including map-projection, were carried out with the freely
1548 available IDL CRISM Analysis Tools (CAT) [Pelkey et al., 2007] and the commercially
1549 available ENVI software package.
1550 CRISM spectra in regions of interest (ROIs) spanning ~100 to 1000 pixels
1551 were averaged to understand large-scale regional mineral assemblages. A spectrally
68 1552 bland or neutral ROI was chosen to represent background martian surface
1553 reflectance properties (e.g. martian dust), and every pixel within an image was
1554 divided by the spectrum corresponding to the neutral ROI within that same image.
1555 This method produces spectral ratios that highlight and accentuate reflectance
1556 properties of ROIs relative to those associated with typical martian dust and/or soil.
1557 Starting with the continuum-removed CRISM images, pixels whose spectra
1558 exceeded a user-defined (and image-dependent) band depth threshold at ~2.205
1559 μm were selected for Gaussian fitting. The threshold used here is band depth
1560 ≥0.005. Though it is not suggested that the same value can be used for every CRISM
1561 image due to differences in noise, for the images used in this study this value is
1562 larger than the background ‘noise’ in the ratio spectra. This was determined by
1563 calculating the mean and standard deviation over the 1.70-1.85 µm wavelength
1564 range (24 channels) for each ratio spectrum (pixel) in each CRISM image. This
1565 wavelength range was chosen because it corresponds to a ‘flat’ portion of the
1566 spectrum for materials in these CRISM images (i.e., there are no strong absorption
1567 features within this range).
1568 Gaussian mean (band center), standard deviation (band width), and
1569 amplitude (band strength) are allowed to vary within user-defined ranges. The
1570 Gaussian fits for at least 10 individual pixels are inspected for each image to ensure
1571 that the model fits are appropriate and residuals values for the spectral fits are
1572 reported for every pixel.
1573
1574 1.2 Field Observations
69 1575 The ASD FieldSpec 3 collects reflected radiation through a fiber optic cable
1576 that is connected to three detectors within the spectrometer housing. The bare fiber
1577 optic cable has a field of view (FOV) of ~25° but can be equipped with different
1578 foreoptics to reduce the FOV to as small as 1°. Field spectra measured at Rio Tinto
1579 using solar radiation were acquired using a 5° FOV foreoptic with a typical standoff
1580 distance of ~1 m from the target of interest. The contact probe uses an internal
1581 quartz-tungsten-halogen (QTH) bulb for illumination and measures a spot size of
1582 ~10 mm. The contact probe has the advantage that viewing/illumination geometry
1583 is fixed and thus identical for all measurements.
1584
1585 1.3 Laboratory Reflectance Measurements
1586 The light source for the laboratory reflectance measurements with the ASD
1587 FieldSpec consisted of an enclosed QTH bulb in combination with a fiber optic cable
1588 that has a 25° FOV. The source fiber was positioned to have a central incidence angle
1589 of 30° (from normal) and the receiving fiber (attached to the spectrometer) was
1590 positioned to have a central emergence angle of 0°. The heights of the fibers were
1591 adjusted as needed in order to fully illuminate and capture the entire sample in the
1592 field of view.
1593
1594 1.4 XRD Measurements and Rietveld Modeling
1595 The D2 Phaser is equipped with a low voltage Cu source and a Lynxeye 1-D
1596 detector. Measurements were made using a 0.6mm divergence slit, 3 mm
1597 antiscatter shield, 2.5˚soller slit and Ni filter and rotating stage (15 rpm).
70 1598 The Rietveld modeling incorporated instrumental parameters included the
1599 Cukα5_Berger emission profile, a 3rd order Chebyshev polynomial (background
1600 modeling), and specimen displacement correction. Strain and crystallite size
1601 broadening were varied to help fit the peak breadths while unit cell parameters and
1602 cation occupancies within each crystal structure were refined to model the
1603 appropriate peak position. A spherical harmonics correction was also applied to
1604 correct for any intensity variations due to preferred orientation during sample
1605 mounting.
Mineral Source Chlorite Zanazzi-et-al.-(2007) Goethite Yang-et-al.-(2006) Gypsum Boeyens-and-Ichharam-(2002) Illite Gaultieri-(2000) Jarosite Basciano-and-Peterson-(2007,-2008);-Scarlett-et-al.-(2010) Muscovite Birgatti-et-al.-(1998);-Birle-and-Tettenhorst-(1968) Pyrite Bayliss-(1977) 1606 Quartz Levien-et-al.-(1980) 1607 S Table 1: Structure file references for the Rietveld refinement. 1608 References: 1609 Basciano, L.C., Peterson, R.C., 2008. Crystal chemistry of the natrojarosite-jarosite and natrojarosite- 1610 hydronium jarosite solid-solution series: A synthetic study with full Fe site occupancy. American 1611 Mineralogist 93, 853–862. doi:10.2138/am.2008.2731 1612 Basciano, L.C., Peterson, R.C., 2007. Jarosite hydronium jarosite solid-solution series with full iron site 1613 occupancy: Mineralogy and crystal chemistry. American Mineralogist 92, 1464–1473. 1614 doi:10.2138/am.2007.2432 1615 Bayliss, P., 1977. Crystal structure refinement of a weakly anisotropic pyrite. American Mineralogist 62, 1616 1168–1172. 1617 Birle, J.D., 1968. Refined Muscovite Structure. Mineralogical Magazine 36, 883–886. 1618 doi:10.1180/minmag.1968.036.282.24 1619 Boeyens, J.C.A., Ichharam, V.V.H., 2002. Redetermination of the crystal structure of calcium sulphate 1620 dihydrate, CaSO4 · 2H2O. Zeitschrift für Kristallographie - New Crystal Structures 217. 1621 doi:10.1524/ncrs.2002.217.jg.9 1622 Brigatti, M.F., Frigieri, P., Luciano, P., 1998. Crystal chemistry of Mg-, Fe-bearing muscovites 2M1. 1623 American Mineralogist 83, 775–785. 1624 Gualtieri, A.F., 2000. Accuracy of XRPD QPA using the combined Rietveld–RIR method. Journal of 1625 Applied Crystallography 33, 267–278. doi:10.1107/S002188989901643X 1626 Levien, L., Prewitt, C.T., Weidner, D.J., 1980. Structure and elastic properties of quartz at pressure. 1627 American Mineralogist 65, 920–930.
71 1628 Ransom, B., Bennett, R.H., Baerwald, R., Shea, K., 1997. TEM study of in situ organic matter on 1629 continental margins: occurrence and the “monolayer” hypothesis. Marine Geology 138, 1–9. 1630 doi:10.1016/S0025-3227(97)00012-1 1631 Scarlett, N.V.Y., Grey, I.E., Brand, H.E.A., 2010. Ordering of iron vacancies in monoclinic jarosites. 1632 American Mineralogist 95, 1590–1593. doi:10.2138/am.2010.3591 1633 Yang, H., Lu, R., Downs, R.T., Costin, G., 2006. Goethite, α-FeO(OH), from single-crystal data. Acta 1634 Crystallographica Section E Structure Reports Online 62, i250–i252. 1635 doi:10.1107/S1600536806047258 1636 Zanazzi, P.F., Montagnoli, M., Nazzareni, S., Comodi, P., 2007. Structural effects of pressure on 1637 monoclinic chlorite: A single-crystal study. American Mineralogist 92, 655–661. 1638 doi:10.2138/am.2007.2341 1639
72